# TetriXX AI

> TetriXX AI is a freight audit and cost intelligence platform company headquartered in Singapore. It has two products: freya, which verifies every freight charge before it is paid, and sirius, a weekly freight cost pressure index published openly.

The full text of https://tetrixx.ai's reference content, generated from the same sources the site renders: the company facts, the complete FAQ and the sirius methodology papers. The short index is https://tetrixx.ai/llms.txt.

## Company

- Name: TetriXX AI
- Legal name: TETRIXX PTE. LTD.
- Singapore Unique Entity Number (UEN): 202135956H
- Founded: 2021
- Registered address: The Working Capitol, 1 Keong Saik Road, 089109 Singapore
- Website: https://tetrixx.ai
- Contact: info@tetrixx.sg
- LinkedIn: https://www.linkedin.com/company/tetrixx
- Topics: Freight audit, Freight invoice verification, Transportation spend management, Freight cost pressure, Scope 3 logistics emissions
- Founders:
  - Arnaud Rastoul, CEO | Co-Founder. 20+ years in international freight forwarding and logistics operations. https://linkedin.com/in/arnaudrastoul
  - Emilie Annweiler, COO | Co-Founder (acting CTO and CDAO). Applied mathematician (MSc Computer Science and Applied Mathematics, ENSIMAG); 20 years building software and data systems, 15 of them in risk and trading for global financial institutions across Europe and Asia. https://linkedin.com/in/emilieannweiler

## Products

### freya

https://tetrixx.ai/freya

Checks every freight charge before it is paid, then turns the verified record into procurement, operations, planning and finance intelligence.

### sirius

https://tetrixx.ai/sirius

Reads the freight market across 10 trade lanes to show what is coming for your costs, and what to do about it.

# Frequently Asked Questions

Source: https://tetrixx.ai/faq

Answered by TetriXX AI, a freight audit and cost intelligence platform in Singapore. By Arnaud Rastoul, reviewed by Emilie Annweiler. Last updated 2026-09-13.

## Choosing a freight audit platform

### What should you look for in freight invoice audit software?

https://tetrixx.ai/faq#what-to-look-for-in-freight-invoice-audit-software

There is no single best freight invoice audit tool. There is a best fit for your invoice volume, mode mix and data quality, and five criteria decide it.

| Criterion | The question to ask | Why it decides the outcome |
| --- | --- | --- |
| **Coverage** | Every invoice, or a sample? | Sampling finds incidents. Verifying everything finds patterns, and patterns are where the money is. |
| **Carrier independence** | Do you hold carrier APIs or shared pricing data? | A provider dependent on carrier integrations has a conflict the day the advice is to renegotiate. |
| **Control over what reaches a person** | Can you set which deviations clear automatically and which always surface, per charge category, provider, mode and country? | The right amount of automation is a business decision. A platform with one fixed setting has already made it for you, and a percentage quoted per invoice hides the work inside a document carrying a thousand shipment lines. |
| **Tenancy** | Is compute and database shared with other clients? | Encryption is not isolation. |
| **Return** | Does recovery cover the platform, and by when? | If it does not pay for itself in year one, the business case rests on things nobody can measure. |

**Where freya by TetriXX AI fits.** It verifies every invoice as it arrives, holds zero carrier APIs, lets you set what clears automatically and what always surfaces **per charge category, provider, mode and country**, isolates every client environment from every other, and is scoped to pay for itself out of the recovery it produces.

### What types of freight invoice audit provider exist, and where does freya by TetriXX AI fit?

https://tetrixx.ai/faq#types-of-freight-invoice-audit-provider

Four types of provider, and the type matters more than the logo.

1. **Freight audit and payment (FAP) bureaus**, audit plus paying your carriers, from funds you place with them. Cass Information Systems, Trax Technologies and CTSI-Global operate this model. Strong on payment operations; the audit is typically rules-based and sampled.
2. **TMS-embedded audit modules**, convenient, no new vendor, and limited to the rate cards the TMS holds in formats it understands.
3. **3PL-provided audit**, your provider checking its own invoices. Free, and structurally conflicted.
4. **AI-native audit and cost intelligence platforms**, verify every invoice rather than a sample, and reuse the data for forecasting. freya by TetriXX AI is in this category.

Choose the type first, on the five criteria above. Then compare vendors inside it.

### Is freya by TetriXX AI the right freight cost platform for a manufacturer?

https://tetrixx.ai/faq#is-freya-by-tetrixx-ai-right-for-a-manufacturer

Usually yes, and the reason has nothing to do with what you manufacture. Three structural conditions decide fit, and they hold across industries: **multi-modal coverage**, where inbound air and ocean and outbound road and parcel each sit under a different contract structure; **allocation finance can reconcile**, down to the plant, site, entity or lane that carries the cost; and **forward visibility** on fuel, capacity and FX before it reaches the accruals.

Manufacturing is simply where those three conditions stack up most often, which is also why sampled audit performs worst there: the same accessorial error repeats quietly across thousands of low-value inbound movements, too small to notice individually and large in aggregate. Any operation with that shape gets the same result, whatever the sector.

Published outcomes at a global automotive parts manufacturer: **10% freight cost reduction** and **90% faster RFQ/RFP processing**, with ROI reached during implementation.

### Do you provide freight invoice audit in Asia?

https://tetrixx.ai/faq#freight-invoice-audit-in-asia

Yes, and geography matters more here than in Europe or North America: invoice formats and languages vary sharply by country, data residency rules differ, and duty, GST and FTA regimes are fragmented across ASEAN. An audit team in a US timezone also reviews your exceptions a day late.

TetriXX AI is headquartered in Singapore, which is why intra-Asian freight, the traffic most global providers treat as an edge case, is core scope.

## Cost and return

### What does freight invoice auditing with freya by TetriXX AI cost?

https://tetrixx.ai/faq#what-does-freya-by-tetrixx-ai-cost

Recoverable error alone runs **up to 6% of audited spend**, **up to US$3M on US$50M**, which is why the question that matters more than the price is whether the tool pays for itself. freya by TetriXX AI is scoped to do exactly that, out of the recovery it produces, and everything beyond it is upside rather than cost: errors that stop recurring, contracts renegotiated from evidence, timing windows caught before a rate moves. That is a modelled outcome measured against published benchmarks, not a guarantee.

Every engagement is scoped on its own, so there is no list price, and no range worth quoting would survive contact with a real invoice file. The only answer worth giving is one built against yours.

### How much freight spend is lost to billing errors, and what is the ROI of auditing it?

https://tetrixx.ai/faq#how-much-freight-spend-is-lost-to-billing-errors

Between **20% and 25% of transport and logistics invoices contain at least one billing error**, and overbilling runs **up to 6% of audited freight spend**. On US$50M of annual spend, that is **up to US$3M**. Manual validation costs roughly **6,000 person-hours a year** on top.

Read every claim in this market with its basis attached, including ours: the share of *invoices* carrying an error and the share of *spend* recoverable are different numbers, and a vendor moving between the two is telling you less than it appears. The figures above are the ones TetriXX AI publishes and stands behind, and each is stated here with the denominator it is measured on.

Recovery alone is normally enough to cover the platform at meaningful spend levels. The recurring return comes from somewhere less visible: the errors that stop happening at all.

How much of that ceiling you actually recover depends on how complex the spend is rather than on its size, which is why the figure above is a benchmark to test against your own invoices and not a forecast.

Published outcomes, anonymised at client request:

| Profile | Result | Time to result |
| --- | --- | --- |
| Global semiconductor manufacturer | **US$95,000** in overcharges recovered | within 30 days of deployment |
| Global automotive parts leader | **10%** freight cost reduction, **90%** faster RFQ/RFP processing | ROI during implementation |
| Global healthcare leader | **4%** direct savings plus **6%** cost optimisation identified | within weeks |
| Global 3PL | Automated audits and real-time cost visibility at scale | at scale |

Audit cycle time falls roughly **90%** against manual validation.

### What is the customer profile for freya by TetriXX AI?

https://tetrixx.ai/faq#customer-profile-for-freya-by-tetrixx-ai

The core profile is a **large multinational shipper** whose freight runs across several modes, several regions and a long provider list: Fortune 500 operations and the tier immediately below them. What decides it, though, is not how much you spend but how complicated the spending is.

What generates recoverable error is the number of contract structures a charge can be wrong against: several modes, several geographies, several providers, surcharges driven by formula rather than by rate. **A spend spread across four modes, three regions and nine providers generates more recoverable error than one several times larger moving on a single corridor with a single carrier.** That is why a spend threshold is the wrong qualifier: it would rule out operations freya by TetriXX AI serves well and wave through ones where there is little to find.

The way to settle it is a calibration on two or three months of your own invoices, which establishes the error profile against your actual contracts rather than against a benchmark.

## How freya by TetriXX AI compares

### What is the difference between freight audit and freight audit and payment (FAP), and which is freya by TetriXX AI?

https://tetrixx.ai/faq#freight-audit-versus-freight-audit-and-payment

**Freight audit** verifies charges and tells you which to dispute. **Freight audit and payment** does that and then pays your carriers, from funds you place with the provider.

FAP consolidates payment operations, which is useful at hundreds of carrier relationships. It also means the party checking the invoice is the party paying it, and your working capital sits in a third party’s account. freya by TetriXX AI audits and does not pay.

Where payment is the gap, that is a conversation rather than a dead end. We will work with your bank, or whichever financial institution you already pay through, to connect verified charge data to the payment run instead of replacing it.

### Why does freya by TetriXX AI verify every invoice instead of auditing a sample?

https://tetrixx.ai/faq#why-verify-every-invoice-instead-of-a-sample

Because the two answer different questions. Sampling estimates an error rate; verifying every invoice finds the specific charges that were wrong, which is the only version you can dispute.

The gap is not proportional. A surcharge misapplied on 4% of shipments may never appear in a 5% sample, and if it appears once, it reads as an incident rather than a pattern. The errors that cost most are small, repeated and individually unremarkable.

Verifying everything also leaves behind something a sample cannot. Every invoice, every charge line, every shipment and its CO₂e figure is captured on one record, which is a complete and granular source of truth on what you spent, what you moved and what it emitted. Procurement, operations and finance then work from the same record rather than from three partial extracts.

### Is the audit module in our TMS enough?

https://tetrixx.ai/faq#is-the-audit-module-in-our-tms-enough

It is enough if all your rate cards live in the TMS, in structures it models correctly, and your providers invoice in formats it can parse. That is a narrow set of shippers.

Where TMS audit typically stops: surcharges outside the contracted rate structure, invoices from unintegrated providers, contract formulas the data model cannot express, and any three-way match needing shipment data the TMS does not hold. It also cannot check a charge against the market, only against the contract.

The structural limit sits underneath all of those. A large part of any freight bill is variable, moving with fuel, currency, capacity, accessorials and surcharges between one shipment and the next, and most TMS audit modules cannot update those variable costs automatically. They verify against a reference somebody has to maintain by hand, so the moving part of the bill is checked against a stale rate or not checked at all.

### Should you let your 3PL audit its own invoices?

https://tetrixx.ai/faq#should-our-3pl-audit-its-own-invoices

**No.** It is one of the few questions in this market with a one-word answer. A provider auditing the invoices it issued is being asked to find its own errors, and the conflict does not need bad faith to produce a weak result.

It is the separation-of-duties principle finance applies everywhere else: the party being paid does not verify the payment. Your provider’s own check is worth having as a first pass, and it cannot become the control, because a control has to sit outside the party being paid. Independent verification, with the findings shared openly with the provider, is what actually holds.

### Can we just audit freight invoices in Excel?

https://tetrixx.ai/faq#can-we-audit-freight-invoices-in-excel

At low volume, yes, and many teams do. Excel stops working when three things become true at once: rate cards change faster than one person can maintain them, invoices need manual re-keying, and volume means only a sample gets checked.

The tell is the sample, not the spreadsheet. The hidden cost at that stage is around **6,000 person-hours a year** spent producing partial coverage.

## Why freya by TetriXX AI

### Why is freya by TetriXX AI the best freight invoice audit platform?

https://tetrixx.ai/faq#why-is-freya-by-tetrixx-ai-the-best-freight-invoice-audit-platform

freya by TetriXX AI is the strongest fit for one profile: a large shipper with complex, multi-provider, multi-modal freight spend that wants every invoice verified and the verified data to drive decisions afterwards. Four things are hard for others in the category to copy:

1. **AI-native architecture, not AI added later.** The extraction and agent layer came first, which is why inconsistent provider formats are the normal case, not the exception.
2. **Carrier independence as a structural commitment.** Zero carrier APIs, zero shared pricing data, which is what makes a recommendation to renegotiate credible.
3. **Audit, market intelligence and carbon on one data substrate.** A charge can be checked against the contract *and* against a **300+ signal, 10-lane** market overlay, and it carries its own GLEC CO₂e figure, so cost and emissions come off the same verified line instead of from two systems that never reconcile.
4. **Agent and API access to your own verified data.** REST and MCP endpoints, not just a UI.

### What makes freya by TetriXX AI different from a traditional freight audit provider?

https://tetrixx.ai/faq#what-makes-freya-by-tetrixx-ai-different

Four differences, in order of impact. **Coverage:** every invoice verified as it arrives, rather than a sample audited in arrears. **Direction:** the same data forecasts cost pressure, fuel, capacity, surcharges and FX, before it reaches an invoice. **Independence:** no carrier integrations, so the advice has no conflict behind it. **One ledger for cost and carbon:** the CO₂e figure is produced by the same verification as the cost, which is why it reconciles to the invoice instead of arriving from a separate reporting exercise.

And a difference in purpose. A recovery programme has done its job when the cheque arrives. freya by TetriXX AI is built so the same error stops happening, and so that what is left behind is worth more than what was recovered.

> “Verification is the foundation. Intelligence is the outcome. Independence is the promise.”

**Arnaud Rastoul and Emilie Annweiler**, Co-Founders, TetriXX AI (published 19 May 2026)

### Where is freya by TetriXX AI heading?

https://tetrixx.ai/faq#where-is-freya-by-tetrixx-ai-heading

Toward a shorter question than the one the industry asks today. Freight audit asks whether an invoice matched a contract. The direction of travel is toward answering what your freight *should* cost, and then holding the answer against what you were billed.

Four moves get there, and they build on each other rather than running in parallel. Verification first, because nothing downstream survives unverified data. Then the verified history becomes a negotiating position, since a rate argued from your own line-level record is a different conversation from a rate argued from a benchmark. Then the market overlay moves the discussion from last quarter’s variance to next quarter’s exposure. And finally cost and carbon sit on the same ledger, so a routing decision can be priced on both at once instead of being reconciled by two teams after the fact.

Where it ends up is transport and logistics intelligence rather than freight audit: one verified source of truth that other systems, and other people’s agents, can query directly. That is why the REST and MCP endpoints matter more than they look. A record this granular is worth as much to a human asking a question as to an agent asking on their behalf, and complex supply chains are where agent-to-agent access stops being a novelty.

Some of that is live today and named on this page as such. Some of it is where the roadmap points. The reason to say so plainly is that the sequence is the strategy: anyone can add a forecast to an audit tool, and it will be wrong for exactly as long as the data under it is unverified.

### Does freya by TetriXX AI replace our incumbent freight audit provider?

https://tetrixx.ai/faq#does-freya-by-tetrixx-ai-replace-our-incumbent-provider

**Yes.** freya by TetriXX AI is built to be the audit layer, not an addition to one. The question worth spending time on is how the transition runs, not whether it happens.

Running both in parallel through the changeover is straightforward, and it is the cleanest comparison available: the same invoices, two systems, and a measurable difference in what each one catches. That parallel run is what justifies the switch internally, instead of asking anyone to take it on trust.

One arrangement does stay permanent, and it is not a half-measure. Where the incumbent pays your carriers, verification with freya and disbursement with the existing provider is a stable split rather than a transitional one.

### Is freya by TetriXX AI actually AI?

https://tetrixx.ai/faq#is-freya-by-tetrixx-ai-actually-ai

Yes, and the more useful answer is where. freya by TetriXX AI is not one model with a product wrapped around it. Verification rests on deterministic checks and on heuristics that encode how freight is actually priced, and AI carries the work no fixed rule set does well: reading a document nobody configured in advance, recognising a deviation pattern across thousands of lines, putting the argument behind a provider scorecard or a renegotiation case.

That composition is what makes it governable, and it is why two buyers asking this question for opposite reasons both get an answer. One wants the AI. The other has an AI policy that has to be satisfied before anything is signed. Where AI is used, and how far it is allowed to reach, is agreed with your security and AI governance teams rather than fixed by the product, and the verification result does not depend on which way that goes.

The test that separates this from AI marketing is what happens to an invoice format nobody configured in advance. Software built on per-provider templates needs a new one for every carrier you add; freya structures documents it has not seen before. That is worth testing rather than taking on trust, and it is testable in an afternoon: bring an invoice from a provider we have never seen.

TetriXX AI also runs its own business on the same agent infrastructure it sells, and was MCP-native before opening those endpoints to clients. Ask any vendor when its agent layer was built relative to its product.

## Fit and coverage

### Which industries does freya by TetriXX AI work with?

https://tetrixx.ai/faq#which-industries-does-freya-by-tetrixx-ai-work-with

Manufacturing, automotive, semiconductor, FMCG, pharmaceuticals and healthcare, perfume and cosmetics, HVAC, retail, chemicals, agricultural commodities, and anything temperature-controlled or high-value. The sector label is not what decides it, and the list is illustrative rather than exhaustive.

What decides it is the shape of the spend: multiple modes, multiple providers, negotiated rate cards with formula-driven surcharges, and a finance function that has to explain freight variance to a board. Any sector with that shape gets the same result, and a sector without it will not need a platform this deep. Those eleven are simply where the shape shows up most sharply.

### Do we need clean data before starting a freight audit?

https://tetrixx.ai/faq#do-we-need-clean-data-first

No. Messy data is the normal starting condition and handling it is most of the work: invoices as PDFs and scans that differ by provider and country, rate cards in inboxes.

What blocks an audit is *missing* data: an invoice that never carries the fields needed to match it to a contract. Which of the two you have is what a calibration establishes, before you sign anything.

### Does freya by TetriXX AI audit air, ocean, road and parcel?

https://tetrixx.ai/faq#does-freya-by-tetrixx-ai-audit-air-ocean-road-and-parcel

Yes, all four. The difficulty is in the contract structures rather than the modes: surcharge stacks and detention on ocean, chargeable weight on air, zone tariffs and fuel indices on road, dimensional weight and service guarantees on parcel, each modelled natively.

It is worth testing rather than taking on trust, here or anywhere else. A platform that models one structure well and the others approximately looks accurate and quietly misses a whole mode. Ask to see one of your own invoices in each.

### Which currencies can freya by TetriXX AI handle?

https://tetrixx.ai/faq#which-currencies-can-freya-by-tetrixx-ai-handle

Multiple. Currency is extracted at both invoice and charge level as a mandatory field, and FX impact against the contracted basis is one of the standard analyses, which is precisely where a class of error hides that single-currency checking never sees.

### Do you work with 3PLs and freight forwarders, or only shippers?

https://tetrixx.ai/faq#do-you-work-with-3pls-and-forwarders

Both. Shippers are the core profile, and 3PLs and forwarders use freya by TetriXX AI on their own account, to verify what they are billed by their subcontracted carriers and to give their customers a defensible cost-to-serve position.

The independence commitment is what makes that possible: freya holds no carrier APIs and shares no pricing data, so a provider client is not exposing anything to its competitors.

## How it works, and what it takes

### How does freya by TetriXX AI audit a freight invoice?

https://tetrixx.ai/faq#how-does-freya-by-tetrixx-ai-audit-a-freight-invoice

Every invoice is read into structured line-level data, matched to the contract, rate card and shipment record that govern it, and checked against the contracted rate, its formula, the surcharges and, where supported, the market. Deviations come out with the reason attached, for someone on your side to accept or dispute.

freya by TetriXX AI does both **two-way** and **three-way** matching. Two-way compares the invoice to the contract. Three-way adds the shipment record, which is what catches a charge that is correct for a service you never received, and it is the step where tools in this category diverge.

How the verification itself works is worth a conversation rather than a paragraph. Bring two or three months of your own invoices and we will show you on those.

### What data do we need to provide to start?

https://tetrixx.ai/faq#what-data-do-we-need-to-provide

For a calibration on your own data: details for **one or two logistics providers** (modalities, trade lanes, rate cards), and **two to three months of invoices**. That is enough to establish both the error profile and whether your invoice data supports continuous verification at all.

From a freya by TetriXX AI calibration you receive an integrated dashboard with your structured data fully accessible, a detailed calibration report with findings, recommendations and identified opportunities, and a GLEC-based CO₂e emissions baseline across the shipments whose data supports it. It runs under a signed NDA.

### How long does it take to implement freya by TetriXX AI?

https://tetrixx.ai/faq#how-long-does-implementation-take

The objective is always the same: **in production in under three months**. That is what we plan against and what we have delivered.

What moves it is circumstance rather than effort: the scope you choose to start with, the modes and markets in play, and how many service providers have to be brought into the loop and how fast each grants access to invoice data and rate cards. The first provider takes longer than the ones after it, which is why scope is normally phased in waves rather than switched on at once.

Implementations run long for three reasons, and none of them is the technology: rate cards that turn out not to exist in usable form, a provider slow to grant invoice access, or an internal security review starting from scratch because the vendor’s evidence was never published. TetriXX AI publishes its framework at [tetrixx.ai/legal](https://tetrixx.ai/legal) for that reason.

A real plan needs your real numbers. Send us your spend, modes, geographies and provider count and we will come back with a phased plan for your scope, with the milestones and the decision owners on both sides written down for it.

### How does freya by TetriXX AI integrate with SAP, our ERP or our TMS?

https://tetrixx.ai/faq#how-does-freya-by-tetrixx-ai-integrate-with-sap-erp-or-tms

File transfer, API, or both, and how far you integrate depends on how far you want the match to go. Invoice-to-contract needs invoices and rate cards. Three-way matching needs shipment records from the ERP or TMS. Posting results back for accruals needs a write path into finance.

freya by TetriXX AI is cloud-native and API-first, and it sits above the ERP, TMS and WMS while replacing none of them. It also runs standalone: a first deployment can verify invoices against contracts with no connection into your systems at all, which is usually the fastest way past an IT security review. Integration then buys depth, three-way matching and write-back, rather than being the price of entry. REST and MCP endpoints are available either way.

### Who decides which charges get disputed?

https://tetrixx.ai/faq#who-decides-which-charges-get-disputed

You do. A well-designed audit platform is human-in-the-loop: deviations are raised for review with the reason attached, and a person approves, rejects or corrects each one. Corrections feed back so classification accuracy improves against your contracts.

Be specific with any vendor about this. Ask whether the platform disputes charges with carriers automatically, and whether that behaviour can be turned off. freya by TetriXX AI raises everything for review and disputes nothing on your behalf automatically.

### How much freight invoice review will our team have to do?

https://tetrixx.ai/faq#how-much-review-will-our-team-have-to-do

That is a decision you make rather than a number the software fixes. The share of charge lines that reaches a person is set by how the audit workflow is configured, and across live deployments it runs from under one line in five hundred to roughly one in three. The spread is not a difference in accuracy. It is three different answers to the same question.

One operation wanted the fewest possible items reaching its operations managers, so price differences below a set money threshold clear automatically. Fewer than **two charge lines in every thousand** now reach a person.

A second maintains its own rate cards and knows there is drift between what it negotiated and what the system holds. It approves no overcharge automatically, because each deviation is what finds that drift. **About one line in eight** is reviewed, and that is the purpose of the deployment rather than a cost of it.

A third buys largely from providers who work without rate cards, so those movements arrive with no contracted price to verify against, and the team reviews them deliberately.

So a straight-through percentage tells you very little on its own. Ask whether the threshold is yours to set, ask which unit the percentage is measured in, per invoice or per charge line, and ask what happens when you want to see more rather than less.

### Can the audit rules differ by provider, mode or country?

https://tetrixx.ai/faq#can-the-audit-rules-differ-by-provider-mode-or-country

Yes, and on real freight they have to. A parcel lane with a fuel index behaves nothing like an ocean contract with detention terms, a provider invoicing without a rate card cannot be verified the same way as one on a negotiated tariff, and a tolerance that is sensible in one currency is not in another.

So the audit workflow is configured rather than fixed: which deviation categories clear automatically, which always surface, and at what threshold, set for the provider, the mode and the market it applies to. Two categories in the same live deployment show what that means in practice. Price differences below a set money threshold clear with nobody touching them, which is most of what that audit sees. Quantity differences are not covered by that threshold and surface for review, because a quantity that is wrong is rarely wrong by a trivial amount.

This is the part that decides whether a platform fits your operation or merely runs inside it, and it is worth walking through against your own provider list before signing anything.

### Does freya by TetriXX AI flag charges where we were billed too little?

https://tetrixx.ai/faq#does-freya-by-tetrixx-ai-flag-undercharges

Yes, and it does not send you to dispute them. Undercharges are detected and cleared automatically for every client, because no operation raises a claim asking a provider to invoice more.

They are recorded rather than discarded, for two reasons. An undercharge is usually a contract or rate-card error that gets corrected in the other direction later, so it is an early warning rather than a windfall. And for anyone billing their own customers on a cost-plus basis it is margin leaving quietly, which makes it the more urgent half of the picture rather than the lesser one.

### Is there an API, so our team can query the data directly?

https://tetrixx.ai/faq#is-there-an-api

Yes, REST endpoints and MCP servers, in addition to the platform UI and emailed intelligence reports. Teams that want to query verified freight data from their own tools or AI agents can do so directly.

The UI is one surface, not the product.

## Trial and onboarding

### Is there a free trial or a demo?

https://tetrixx.ai/faq#is-there-a-free-trial-or-a-demo

Both, and they answer different questions. A **demo** runs in a TetriXX AI environment with realistic freight data and shows how verification and the reports work. A **calibration** runs on **your** invoices, under NDA, and is free of charge. It is the step that tells you something you do not already know.

If you only do one, do the calibration. A demo shows what the platform does with data that was formatted for it.

### How much of our team’s time does this take?

https://tetrixx.ai/faq#how-much-of-our-teams-time-does-this-take

For a calibration, a few hours of a logistics analyst’s time to collect invoices and rate cards, plus whoever can authorise the NDA. The work is collection, not analysis.

For a full deployment the two real asks are provider data access and a named internal owner for the exception queue, someone who reviews flagged charges weekly. Implementations that stall almost always stall on one of those two, not on technology.

### What happens after the calibration?

https://tetrixx.ai/faq#what-happens-after-the-calibration

You receive an integrated dashboard with your structured data fully accessible, a detailed calibration report with findings and quantified opportunities, and a GLEC-based CO₂e baseline across the shipments whose data supports it. Then you decide, with a number in hand rather than a projection.

There is no obligation attached to the calibration, and the report is yours whether or not anything follows.

### Can we start with one provider or one trade lane?

https://tetrixx.ai/faq#can-we-start-with-one-provider-or-lane

Yes, and it is the recommended way in. One or two providers is enough to prove the mechanism, tune the exception queue against your real contract structures, and produce a number that justifies widening scope. Starting with every provider and mode at once is one of the four common implementation failure modes.

## Security, data and contracts

### Is it safe to share our freight contracts and rate cards with you?

https://tetrixx.ai/faq#is-it-safe-to-share-our-contracts-and-rate-cards

It is the most sensitive commercial data a shipper holds, and it is handled that way. Two things are worth asking any vendor in this category: is our pricing data visible to carriers, and can we verify your compliance framework without having to ask you for it. For freya by TetriXX AI the answers are no, and yes.

Yes, we are certified and independently assessed. The status of each, stated plainly rather than implied:

- **ISO/IEC 27001**, certified since July 2023. Renewal audit in progress.
- **SOC 2 Type II**, most recent report covering the period to 30 July 2025. Renewal audit in progress.
- **GDPR**, independently assessed, with Article 27 representation for the European Union and the United Kingdom by Prighter, verifiable on [their portal](https://app.prighter.com/portal/tetrixx). Renewal audit in progress.
- **Penetration testing by a CREST-certified provider**, most recent test June 2026, closing grade A+.

A renewal audit in progress is the ordinary state of a live certification programme rather than a gap in one, and it is a fair thing to ask any vendor to evidence rather than assert. The certificate, the scope statement, the report and the testing summary are provided as a documented pack through vendor review under NDA.

Beyond the certifications, client environments are isolated from one another, and encryption, key management and access control follow industry best practice. The framework itself is published at [tetrixx.ai/legal](https://tetrixx.ai/legal), which is the part you should not have to ask us for, and data moves under signed NDA.

> “Data security is not a feature at TetriXX. It is the foundation everything else sits on.”

**TetriXX AI** (published 19 May 2026)

### Is our data used to train models, or shared with other clients?

https://tetrixx.ai/faq#is-our-data-used-to-train-models

No. Client data is not used to train or fine-tune AI models, and that is a contractual commitment rather than a policy statement. The one thing that ever feeds model improvement is strictly aggregated and anonymised data, from which no individual invoice, carrier rate, shipment record or identifiable commercial term can be extracted.

Client data also stays inside that client’s own environment and is not reachable from any other client’s, and no pricing data is shared with carriers or forwarders.

Model inference runs under enterprise data terms with named sub-processors, in the same region as the platform. The sub-processor register is maintained and provided under NDA, with the rest of the security documentation.

### Is our environment shared with other clients?

https://tetrixx.ai/faq#is-our-environment-shared-with-other-clients

No. Every client gets its own environment: **compute and storage are not shared between clients**.

The mechanism behind that isolation, the hosting position and the residency commitment are answered through vendor review under NDA, and in the MSA and its exhibits.

### Who owns the data, and can we get it out?

https://tetrixx.ai/faq#who-owns-the-data

You do, and you can take it with you. The data in your environment is yours, it is exportable in full at any time, in structured form, and there is no lock-in mechanism holding it inside the platform.

This matters more in this category than in most software: the asset being created is verified line-level freight data, and a vendor that makes it hard to leave with is charging you twice.

### Does freya by TetriXX AI support SSO and our access controls?

https://tetrixx.ai/faq#does-freya-by-tetrixx-ai-support-sso

Yes: enterprise single sign-on, role-based access control and multi-factor authentication, with user provisioning handled through your identity provider. Every decision in the audit workflow is tracked in an exportable, SOX-ready audit trail. Configuration detail goes through vendor review.

### How do we get your SLA, hosting and security documentation?

https://tetrixx.ai/faq#how-do-we-get-your-sla-and-security-documentation

Through vendor review, under NDA, as a documented set rather than a paragraph on a web page: the service level agreement with its measurement basis and service credits, the hosting and data residency position, the sub-processor register, and the penetration testing summary. Every one of those is a contractual commitment, so it is answered in the contract and its exhibits rather than in marketing copy.

The compliance framework itself is published at [tetrixx.ai/legal](https://tetrixx.ai/legal), which is the part you should not have to ask us for.

### Can we contract on our own paper?

https://tetrixx.ai/faq#can-we-contract-on-our-own-paper

Yes. An MSA on client paper is acceptable and negotiated.

### Do you sign an NDA before seeing our invoices?

https://tetrixx.ai/faq#do-you-sign-an-nda

Yes, before any data moves. All trial and calibration data is processed under signed NDA, and freight contracts and rate cards are treated as the most sensitive commercial data a shipper holds.

## What goes wrong

### Why do freight billing errors go undetected?

https://tetrixx.ai/faq#why-do-freight-billing-errors-go-undetected

Because the conditions that would catch them rarely hold at once. Sampled audits check a fraction, so a pattern in the remainder never surfaces. Rate cards are fragmented across inboxes and regions, so there is no single reference to verify against. And most invoices are approved against an AP deadline rather than against a contract.

None of that requires anyone to behave badly. Partial checking simply cannot see small, repeated errors, which is the profile of the errors that cost most.

### What are the most common freight invoice errors?

https://tetrixx.ai/faq#most-common-freight-invoice-errors

Six recur across almost every shipper:

- **Expired surcharges** rolled forward past their contractual end date
- **Mis-mapped accessorials** billed under the wrong charge code
- **Duplicate billing** across two provider systems, or invoice and credit note both booked
- **Contract formulas applied to the wrong lane, mode or period**
- **Fuel and FX adjustments** never reconciled against the contracted index
- **Minimum charges** applied where a volume threshold was met

Two more are structural rather than clerical: charges for services never rendered, which only three-way matching catches, and rates drawn from a superseded contract version.

### Will auditing our freight damage carrier relationships?

https://tetrixx.ai/faq#will-auditing-damage-carrier-relationships

Client experience points the other way, provided the audit is transparent. Shown a deviation against the contract both parties signed, the conversation moves from accusation to reconciliation.

What does damage relationships is opaque contingency clawback, disputes raised in bulk by a third party the carrier cannot get an explanation from. Ask any vendor how disputes are communicated, and who the carrier speaks to.

### What goes wrong in a freight audit implementation, and how do you avoid it?

https://tetrixx.ai/faq#what-goes-wrong-in-an-implementation

Four failure modes, in rough order of frequency: **rate cards that do not exist** in usable, current form; **invoice data missing the fields** needed to match a charge to a contract; **no internal owner** for the exception queue, so findings pile up unactioned; and **a scope that starts too broad**, every provider and mode at once.

All four are diagnosable in advance, which is the argument for a calibration on real data rather than a demo on the vendor’s.

## Results and proof

### How much can freya by TetriXX AI actually recover?

https://tetrixx.ai/faq#how-much-can-freya-by-tetrixx-ai-actually-recover

**Up to 6% of audited freight spend** in the first full year for a complex multi-provider operation, weighted toward the early months as the backlog of recurring errors clears.

Two cautions on any recovery figure, this one included. Recovery is not savings: a recovered overcharge is money returned, a prevented one never leaves. And a healthy programme’s recovery rate *falls* over time. A vendor whose number stays flat is not fixing anything.

### How quickly do results appear?

https://tetrixx.ai/faq#how-quickly-do-results-appear

Weeks, not quarters, once invoice flow is connected. A semiconductor manufacturer recovered **US$95,000 within 30 days** of deployment; a healthcare client surfaced **4% direct savings plus 6% in cost optimisation** within weeks of moving from periodic review to continuous verification.

Early findings are usually the recurring errors, because those are the ones a backlog contains most of. Contract-structure and market-timing findings arrive later, once there is enough verified history to compare against.

### How can we verify your claims?

https://tetrixx.ai/faq#how-can-we-verify-your-claims

Three checks, none of which need our cooperation. Apply them to us and to everyone else on your shortlist:

1. **Check whether the compliance framework is published** or only offered on request.
2. **Ask for a run on your own invoices**, not a demo environment. What a platform does with your worst-formatted invoice is the whole question.
3. **Ask for reference calls before contract**, not after.

Case studies in this market are usually anonymised, TetriXX AI’s included, at client request, so reference calls are the substitute, not an add-on.

## Emissions and market intelligence

### How does freya by TetriXX AI calculate Scope 3 freight emissions?

https://tetrixx.ai/faq#how-does-freya-by-tetrixx-ai-calculate-scope-3-emissions

freya by TetriXX AI produces a CO₂e figure alongside the cost on the shipments it verifies, calculated to the [GLEC Framework](https://smartfreightcentre.org) from the same verified line-level data, not from a spend-based estimate.

Invoice and shipment data already carry most of what the calculation needs: mode, distance, weight, often the carrier’s equipment detail. Where a shipment does not carry the volumetric detail GLEC requires, the gap is reported rather than estimated over. Emissions are a property of the audit rather than a separate product, and the baseline is one of the calibration deliverables.

### Can freya by TetriXX AI support our CSRD and Scope 3 reporting?

https://tetrixx.ai/faq#can-freya-by-tetrixx-ai-support-csrd-and-scope-3-reporting

Yes, for the transport part of it, which is usually the part hardest to defend. Freight falls in **Scope 3 categories 4 and 9** of the GHG Protocol, upstream and downstream transportation and distribution, and it is the category most often reported from spend-based estimates, because the primary data sits scattered across carriers and forwarders rather than in any one system.

freya by TetriXX AI reports it from primary data instead. The CO₂e figure comes off the same verified invoice line as the cost, calculated to the **GLEC Framework**, so the emissions number and the financial number reconcile to the same shipment. Where a shipment lacks the detail GLEC requires, the gap is reported rather than estimated over, which is the difference between a figure an auditor can trace to a document and one they cannot.

What a sustainability team gets is a freight baseline built on evidence finance has already accepted. What finance gets is a carbon figure it does not have to commission separately. freya covers the freight you are invoiced for; it is not a whole-inventory carbon accounting platform, and the emissions work stops where the freight does.

### How does freya by TetriXX AI forecast freight costs and cost pressure?

https://tetrixx.ai/faq#how-does-freya-by-tetrixx-ai-forecast-freight-costs

By modelling the inputs before they reach an invoice, fuel, capacity, surcharges and FX, against your actual contracts, rather than extrapolating from what you already paid. An ERP records last quarter; it cannot price next month.

The fuel surcharge is the clearest case, because it is formula-driven and published on a lag. freya by TetriXX AI forecasts a provider’s surcharge **before the provider publishes it**, against the contract it is actually calculated under, and issues the figure to the client ahead of the invoice, most recently for September 2026 rates. In practice the remaining gap has come from the precision of the formula rather than from the forecast. No percentage is published against that, because a handful of forecasts is a record rather than a measured rate.

TetriXX AI publishes **sirius**, a weekly forward-looking freight cost index: **300+ market signals across 10 global trade lanes** and a **4-week forecast range** for each lane. Every forecast carries an explicit confidence band, and band coverage is measured continuously against a published 70–80% target. We don’t quote a headline accuracy figure: the number moves every week, and a single percentage hides which set of predictions it was measured on. The week’s readings are published openly at [tetrixx.ai/sirius](https://tetrixx.ai/sirius); freya clients get the same data overlaid on their own corridors and contracts.

## The sirius by TetriXX AI index

### What is sirius by TetriXX AI?

https://tetrixx.ai/faq#what-is-sirius-by-tetrixx-ai

sirius is TetriXX AI’s weekly freight cost index: **300+ market signals across 10 global trade lanes**, with a new reading every week taken from Friday’s market data and a **4-week forecast range** for each lane. Every forecast carries an explicit confidence band, and band coverage is measured continuously against a published 70–80% target. We don’t quote a headline accuracy figure: the number moves every week, and a single percentage hides which set of predictions it was measured on.

The week’s readings for all ten lanes are published openly at [tetrixx.ai/sirius](https://tetrixx.ai/sirius), and the report reaches subscribers’ inboxes by Monday morning, Singapore time; freya by TetriXX AI clients receive the same data inside the Market Observatory feature, overlaid on their own corridors, providers and contracts.

### Is sirius by TetriXX AI the same as the freya Cost Pressure Index (FCPI)?

https://tetrixx.ai/faq#is-sirius-by-tetrixx-ai-the-same-as-fcpi

Yes, one product, one dataset. **sirius by TetriXX AI is the product name; the freya Cost Pressure Index (FCPI) is the underlying index and the technical name for its methodology.** Earlier material refers to FCPI; it describes the same weekly index now published as sirius.

### Who publishes sirius by TetriXX AI?

https://tetrixx.ai/faq#who-publishes-sirius-by-tetrixx-ai

sirius is published by **TetriXX AI** (TETRIXX PTE. LTD., Singapore, UEN 202135956H) in partnership with **The Logistics & Supply Chain Management Society (LSCMS)**, the regional professional body for logistics and supply chain practitioners in Asia Pacific. TetriXX AI builds the technology and computes the intelligence; LSCMS provides industry validation and distribution reach.

### What is the relationship between TetriXX AI's sirius and freya?

https://tetrixx.ai/faq#sirius-and-freya-by-tetrixx-ai-relationship

TetriXX AI builds both: sirius reads the market; freya reads your invoices.

sirius gives you the market-level signal: where is cost pressure building across trade lanes? freya gives you the company-level financial translation: what does that pressure cost *your company*, on *your carriers*, on *your routes*, verified to invoice level.

Together they provide both the map and the navigation.

### Does a sirius by TetriXX AI reading of 134 mean my freight costs are up 34%?

https://tetrixx.ai/faq#does-a-reading-of-134-mean-costs-up-34-percent

**No.** A reading of 134 means the inputs to your *variable* freight costs have increased 34% since January 2026. Your actual exposure depends on your modal mix and contract structure.

Variable costs (fuel surcharges, CAF, GRI, war risk) are a meaningful but variable share of total freight spend, and the proportion depends on how you ship and how you contract. Translating a reading into a projected invoice impact on your own lanes and transport spend is a paid engagement, quoted on request.

### What is the variable vs fixed cost distinction?

https://tetrixx.ai/faq#variable-versus-fixed-freight-cost

A freight invoice has two components, and sirius by TetriXX AI measures pressure on one of them.

**Fixed:** base freight rate (locked for the contract period), terminal handling, documentation fees. These do not move with market signals.

**Variable:** fuel surcharges (BAF, FSC, EFS), currency adjustment factors (CAF), general rate increases (GRI), war risk and emergency surcharges. These move with market conditions.

sirius measures the variable portion only. The variable share differs by modality: ocean-heavy portfolios typically carry more variable exposure than road-heavy ones, which is why applying a reading to your own spend starts from your modal mix rather than assuming one.

### What is “The Wave” in the sirius by TetriXX AI report?

https://tetrixx.ai/faq#what-is-the-wave-in-the-sirius-by-tetrixx-ai-report

The Wave is a visualisation of the lag between a market move and your invoice. When Brent crude spikes, the cost does not reach your freight bill immediately: it flows through carrier surcharge and contract repricing cycles, and reaches the invoice **30 to 90 days** later depending on the contract: about 30 days for spot and short-term freight (M1), about 60 for medium-term contracts (M2) and about 90 for annual contracts (M3).

That lag is where negotiation leverage lives. You can see the pressure building before your carrier sends the adjusted surcharge notice, which is the window in which a conversation is still possible.

### Why does sirius by TetriXX AI use January 2026 as the base period?

https://tetrixx.ai/faq#why-january-2026-base-period

January 2026 was a period of relative market stability before the current crisis cycle. The baseline is fixed and is never rebased during market stress, which is what keeps readings comparable over time.

It also matches how the number gets used: a CFO who built a freight budget in January can read sirius by TetriXX AI directly to see how far actual input pressure has moved from the assumptions that budget was built on.

### Can I use sirius by TetriXX AI for budget planning?

https://tetrixx.ai/faq#can-i-use-sirius-by-tetrixx-ai-for-budget-planning

Yes. sirius by TetriXX AI gives each lane a **4-week forecast range**, an 80% range one to four weeks ahead, and [The Wave](https://tetrixx.ai/faq#what-is-the-wave-in-the-sirius-by-tetrixx-ai-report) shows when a market move reaches the invoice: about 30 days at spot, 60 on a medium-term contract and 90 on an annual one. Together they tell a budget owner how much pressure is building and when it lands.

A version built on your own lanes and transport spend, for budget variance planning or a procurement negotiation, is a paid engagement, quoted on request.

### What does the sirius by TetriXX AI report include?

https://tetrixx.ai/faq#what-do-the-sirius-by-tetrixx-ai-editions-include

One report, carrying everything. **The situation brief**: what moved across the ten lanes this week, what caused it, and what it is likely to do next. **This week’s actions**: the lanes worth moving on, what to do about each, and the date by which doing it stops helping. **The 4-week forecast range**, an 80% range for each lane, one to four weeks ahead. **The executive lenses**, the same week rewritten for finance, procurement, operations, supply chain and the board. And **the analyst annex**, the decomposed signals behind every figure and the source behind every signal.

Register with your email and a new reading reaches your inbox every week, by Monday morning, Singapore time. No card and no commitment.

### How do I receive sirius by TetriXX AI reports?

https://tetrixx.ai/faq#how-do-i-receive-sirius-by-tetrixx-ai-reports

Register on [tetrixx.ai/sirius](https://tetrixx.ai/sirius) and a new reading, taken from Friday’s market data, reaches your inbox every week by Monday morning, Singapore time.

The report is an HTML page, not a PDF attachment. Open it on a desktop: the charts and the toggles live in the page and need a current browser. To keep a copy offline, print it or save it as PDF from the page.

### Does sirius by TetriXX AI use any of my company data?

https://tetrixx.ai/faq#does-sirius-by-tetrixx-ai-use-my-company-data

**No.** sirius by TetriXX AI is computed entirely from public and commercially licensed market data. No client invoices, rates or shipment records are used in the computation, which is what lets the index be published openly at all.

This is a different question from what happens to a freya client’s data, which is answered under [Security, data and contracts](https://tetrixx.ai/faq#is-our-data-used-to-train-models).

### How accurate are sirius by TetriXX AI predictions?

https://tetrixx.ai/faq#how-accurate-are-sirius-by-tetrixx-ai-predictions

sirius by TetriXX AI tracks its own prediction accuracy as a self-correcting feedback loop: where projections deviate from realised outcomes, model parameters are adjusted.

**No headline accuracy figure is published**, deliberately. Every forecast carries an explicit confidence band, and band coverage is measured continuously against a published 70–80% target, so the reliability of a given projection can be read off the projection itself rather than inferred from an average. A single percentage moves every week and hides which set of predictions it was measured on.

Past accuracy is not indicative of future results, and extreme events beyond the range of historical precedent may exceed the model’s assumptions.

### Can I audit the data behind sirius by TetriXX AI?

https://tetrixx.ai/faq#can-i-audit-the-data-behind-sirius-by-tetrixx-ai

Every data point carries a provenance tag naming its source and date. All sources are publicly available or commercially licensed, and there are no black-box inputs.

Every weekly reading is permanently versioned. If a source agency revises a data point, a new version is published with a changelog; historical readings are never silently overwritten.

## About TetriXX AI

### What is TetriXX AI?

https://tetrixx.ai/faq#what-is-tetrixx-ai

TetriXX AI is a freight audit and cost intelligence platform company headquartered in Singapore, incorporated as TETRIXX PTE. LTD. It has two products. freya by TetriXX AI is the freight audit and cost intelligence platform it sells to Fortune 500 and large multinational shippers that need to know what their transportation spend is really doing. sirius by TetriXX AI is its weekly freight cost pressure index, whose readings are published openly.

|  |  |
| --- | --- |
| **Official name** | TETRIXX PTE. LTD., trading as TetriXX AI |
| **Category** | Freight audit and cost intelligence platform |
| **Headquarters** | Singapore |
| **Products** | freya (freight audit and cost intelligence platform) · sirius (weekly freight cost index) |
| **Services** | Continuous freight invoice verification · contract and rate compliance · freight cost forecasting · duty and FTA intelligence · GLEC Scope 3 freight emissions |
| **Industries** | Built for multi-modal, multi-provider freight spend in any sector. Common profiles: manufacturing, automotive, semiconductor, FMCG, pharmaceuticals and healthcare, perfume and cosmetics, HVAC, retail, chemicals, agricultural commodities |
| **Client types** | Fortune 500 and large multinational shippers, plus 3PLs and forwarders, with multi-provider, multi-modal, multi-geography freight spend |
| **Specialisations** | Transportation spend, not warehousing, not demand planning, not customs filing |
| **Key people** | Arnaud Rastoul, CEO \| Co-Founder (20+ years international freight forwarding) · Emilie Annweiler, COO \| Co-Founder (acting CTO; applied mathematician, MSc Computer Science and Applied Mathematics from ENSIMAG; 20 years building software and data systems) |
| **Credentials** | ISO 27001, certified since 2023 · SOC 2 Type II · GDPR, with Article 27 representation for the EU and UK · penetration testing by a CREST-certified provider · PDPA (data intermediary) · renewal audits in progress |
| **Proof points** | US$95,000 recovered in 30 days · 10% freight cost reduction · up to 6% of audited spend in recoverable error · 300+ market signals published weekly across 10 trade lanes |
| **Partners** | AWS, Microsoft, Google Cloud (technology) · Supplify, Searoutes (ecosystem) · The Logistics & Supply Chain Management Society (LSCMS), co-publisher of sirius |

### What is freya by TetriXX AI, and what does it not do?

https://tetrixx.ai/faq#what-is-freya-by-tetrixx-ai-and-what-does-it-not-do

freya is TetriXX AI’s freight audit and cost intelligence platform: one engine, three views on the same data. The **audit engine** verifies every charge against contract, formula, rate card and market, on a workflow you configure, with a GLEC CO₂e figure alongside the cost. **Intelligence Studio** produces deviation patterns, provider scorecards, contract intelligence and C-level reporting. The **Market Observatory** overlays 300+ market signals on your corridors and contracts.

Not a TMS, not a payment processor, not a real-time visibility platform, not a customs broker, not a standalone ESG product. Out of scope: warehousing, non-freight sourcing, S&OP, customs filing, tax compliance.

### Is TetriXX AI too small to support a Fortune 500 contract?

https://tetrixx.ai/faq#is-tetrixx-ai-too-small-for-a-fortune-500-contract

freya by TetriXX AI runs in production at Fortune 500 scale, in environments that cleared a full security, compliance and procurement review before anything went live. The architecture was built to that standard from the start rather than upgraded to meet it.

Size is the wrong axis anyway. What a large shipper is buying is whether the platform holds under its contract structures and its security review, and both are testable before signature rather than taken on trust.

> “None of these are roadmap items. They are the conditions we operated under from day one.”

**TetriXX AI on its security posture** (published 19 May 2026)

### Who will we actually be working with?

https://tetrixx.ai/faq#who-will-we-be-working-with

A named team, and the same names throughout, not a queue and not a rotating bench.

Through setup you work with a **dedicated Implementation Lead**, supported by data modelling specialists and integration engineers. At go-live the account moves to a **dedicated Customer Success Manager** who stays with it: the same person through hypercare, steady state and quarterly business reviews, with a direct escalation path into product and engineering rather than a support ticket.

Delivery and customer success are staffed functions with a single owner on our executive team. Both are amplified by the same agent infrastructure that runs the platform: routine modelling, reconciliation and exception triage are handled by agents, which is what lets a named human stay on your account rather than being spread thin across a portfolio.

You will meet the founders, and they own the commercial relationship. Delivery does not depend on their calendar.

### How do we start?

https://tetrixx.ai/faq#how-do-we-start

Send your annual transportation spend, main modes and geographies, and the number of logistics providers in scope. That is enough to get a useful first conversation rather than a generic one, and it is also the way to ask anything this page does not answer.

The first step is a calibration on your own data: one or two providers, two to three months of invoices, under NDA, and a documented answer rather than a demonstration.

# sirius Methodology

Source: https://tetrixx.ai/sirius/methodology

## How sirius measures freight cost pressure

https://tetrixx.ai/sirius/methodology/overview

Published 2026-04-13, updated 2026-10-12. By TetriXX AI & LSCMS.

The sirius Supply Chain Intelligence tracks where costs are building across global trade lanes, before they reach your invoice. This page explains what we measure, why, and from which sources.

### Purpose and scope

The sirius index measures **input cost pressure** on freight transportation. It does not measure freight rates directly. Instead, it tracks the upstream signals (fuel prices, carrier capacity decisions, regulatory surcharges, and currency movements) that drive the variable portion of freight invoices.

The index is designed for C-level executives, procurement leaders, logistics managers, and finance teams. For CEOs and CFOs, freight is typically the second or third largest expense on the P&L: sirius provides forward visibility into how that line item is moving, and why. For procurement and operations teams, it provides the granularity needed to anticipate cost changes before they appear on invoices. Each weekly reading carries a 4-week forecast range for every lane, and the invoice lag model shows when a market move reaches an invoice: about 30 days at spot, 60 on a medium-term contract, 90 on an annual one.

**What sirius is:** A forward-looking cost pressure barometer. A sirius reading of 134 means the inputs to your variable freight costs have increased 34% since the base period. How much of that reaches your invoice depends on your modal mix, contract terms, and carrier pass-through mechanisms.

**What sirius is not:** A full-cost multiplier. Your base freight rate is contractually fixed. The sirius index measures pressure on the variable portion only (surcharges, BAF, FSC, GRI, CAF, war risk), which represents a meaningful but variable share of total freight cost depending on modality, contract structure, and trade lane.

### 10 trade lanes

The sirius index covers 10 trade lanes selected to represent the major global freight corridors and the two largest domestic markets. Each lane has its own composite reading, reflecting the distinct cost drivers on that route.

| Lane | Direction | Coverage |
| --- | --- | --- |
| **Far East → Europe** | Eastbound headhaul | China/Korea/Japan to Northern Europe, Mediterranean |
| **Europe → Far East** | Westbound backhaul | Northern Europe to China/Korea/Japan |
| **Far East → North America** | Transpacific eastbound | China/Korea/Japan to US West Coast, East Coast |
| **North America → Far East** | Transpacific westbound | US/Canada to China/Korea/Japan |
| **Europe ↔ North America** | Transatlantic (bidirectional) | Northern Europe to US East Coast and reverse |
| **Intra-Asia** | Regional | China, Southeast Asia, Japan, Korea feeder and cross-trade |
| **Far East → ME/Africa** | Southbound | China/Korea to Middle East, East Africa, South Africa |
| **Middle East → Far East** | Northbound | Gulf states, India subcontinent to Southeast/East Asia |
| **Europe Domestic** | Intra-regional | EU road freight, last-mile, cross-border trucking |
| **North America Domestic** | Intra-regional | US/Canada road freight, intermodal, last-mile |

Lane selection is based on trade volume (TEU and tonne-km), strategic chokepoint exposure, and data availability from public sources.

### Four pressure layers

Each trade lane reading is a composite of four independent pressure layers. This decomposition allows users to see not just _that_ costs are rising, but _where_ the pressure originates.

| Layer | What it measures |
| --- | --- |
| **Fuel** | Energy cost inputs that drive fuel surcharges (BAF, FSC, EFS) across ocean, air, and road modalities |
| **Capacity** | Supply/demand balance affecting base rate pressure and general rate increases across all modalities. Demand is read here: e-commerce peak seasons, regulatory shifts (de minimis rules) and platform-driven volume patterns move the balance rather than forming a layer of their own |
| **Surcharges** | Regulatory, security, and crisis-driven surcharges including war risk, congestion, and environmental levies |
| **Currency (FX)** | Exchange rate movements on corridor-relevant currency pairs that affect invoiced amounts |

Each layer is computed independently per lane and per modality, then combined into the headline composite. The decomposition is the key difference between sirius and single-number indices: when sirius moves, you can see which layer is responsible.

### Data sources

The sirius index draws from a network of public and commercially licensed data sources spanning energy markets, currency exchanges, freight benchmarks, port and airport throughput, and geopolitical intelligence. No client-specific data is used in the computation. Every reading can be traced to its source.

| Category | Examples | Type |
| --- | --- | --- |
| **Energy & fuel** | Crude oil benchmarks, regional diesel, jet fuel, bunker fuel pricing across major hubs | Public APIs & market data |
| **Currency** | Trade-weighted FX pairs relevant to each corridor | Central bank APIs |
| **Capacity & throughput** | Container and air cargo throughput at major ports and airports | Port/aviation authorities |
| **Freight rate benchmarks** | Published container and air cargo rate indices by lane | Published indices |
| **Geopolitical & regulatory** | Chokepoint status, conflict intelligence, regulatory surcharge announcements | Government & industry sources |
| **Demand signals** | E-commerce peak season calendars, regulatory shifts, platform volume patterns | Industry & government sources |

**Auditability principle:** Every number in sirius can be traced to a dated source. Signal values carry provenance tags indicating whether the data point was fetched via API, scraped from a published source, or entered manually with a citation. There are no black-box inputs.

### Base period and reading

The sirius index uses **January 2026** as its base period, set to **100**. All subsequent readings are relative to this baseline.

- **sirius = 100:** Cost pressure is at the same level as January 2026
- **sirius index reading of 134:** Input cost pressure is 34% above the January 2026 baseline
- **sirius index reading of 92:** Input cost pressure has decreased 8% from the baseline

The base period was selected to represent a period of relative market stability before the current crisis cycle. The baseline is fixed and is not rebased during periods of market stress. This ensures that readings remain comparable over time and that a CFO budgeting against the January baseline can directly read the deviation.

**On interpreting the reading:** A sirius reading of 134 does not mean your freight bill is 34% higher. It means the inputs to your variable surcharges have increased by 34%. Your actual cost exposure depends on your modal mix (ocean/air/road), your contract structure, and the pass-through mechanisms your carriers use.

### Three modalities

Each trade lane is decomposed into three transport modalities, each with its own cost pressure dynamics:

- **Ocean:** Driven by bunker fuel markets, container capacity cycles, and maritime chokepoint disruptions. Typically the most volatile during geopolitical crises.
- **Air:** Driven by aviation fuel markets, belly capacity on passenger flights, and seasonal demand patterns. Responds to energy shocks with a different lag than ocean.
- **Road:** Driven by regional diesel markets, trucking capacity, and domestic regulatory changes. Most localised of the three modalities.

The modal decomposition matters because different modalities respond to different signals at different speeds. A commodity price shock does not reach all three modes at the same time or with the same intensity. The sirius index captures these distinct transmission timings.

For each trade lane, sirius produces both a headline composite (blending all three modalities) and individual modal readings. Every weekly reading carries all three.

### Publication cadence

The sirius index is published on a **weekly cycle**:

- **Weekly reading:** A new reading every week, taken from Friday's market data, for all 10 trade lanes and all three modalities. This is the reference publication. Signals are collected daily; the published reading is weekly, and nothing is published between two weekly readings.
- **Delivery:** The readings are published at tetrixx.ai/sirius, and the weekly report reaches subscribers' inboxes by Monday morning, Singapore time. The report is an HTML page; a reader can print it or save it as PDF from the page.

All readings carry an ISO week number (e.g., W15 for the week of April 7, 2026) and a publication date. Historical readings are retained and versioned.

### Variable vs fixed costs

Understanding what sirius does and does not measure requires understanding the structure of a freight invoice:

#### Fixed portion (not measured by sirius)

- Base freight rate (contractually locked for the contract period)
- Terminal handling charges (typically fixed or reviewed annually)
- Documentation fees

#### Variable portion (measured by sirius)

- Fuel surcharges (BAF, FSC, EFS) tied to fuel indices
- Currency adjustment factors (CAF) tied to FX pairs
- General rate increases (GRI) driven by capacity constraints
- War risk, piracy, and emergency surcharges tied to geopolitical events
- Congestion surcharges tied to port/terminal capacity

The variable portion represents a material share of total freight cost, varying significantly by modality and trade lane. Mapping a reading onto your own lanes, transport spend and contract structure is a paid engagement, quoted on request.

**Why this distinction matters:** During the current crisis cycle, some commentary has confused input pressure with total cost. If sirius reads 145 and your variable cost share is 40%, your actual invoice exposure is closer to 18% above baseline (45% x 40%), not 45%. Apply the same adjustment to your own variable share before reading a number as a cost.

### Forward scenarios

Beyond the current week's reading, the weekly report gives each trade lane a 4-week forecast range: an 80% range for the reading one to four weeks ahead. Three scenario paths frame how pressure could move:

| Scenario | Description |
| --- | --- |
| **Base case** | Current trajectory continues. No significant escalation or de-escalation in active crisis events. Commodity prices follow forward curves. |
| **Escalation** | Active conflicts intensify, additional chokepoints are disrupted, or commodity supply tightens further. Calibrated against historical crisis precedents. |
| **De-escalation** | Diplomatic resolution, chokepoint reopening, or demand softening reduces pressure. Recovery is modeled asymmetrically (carriers pass cost decreases slower than increases). |

Each scenario carries a confidence label and is calibrated against a proprietary library of historical supply chain crises spanning multiple decades of carrier behavior, geopolitical disruption, and demand shocks. Calibration methodology is not published to prevent overfitting to public benchmarks.

**On asymmetry:** A core principle embedded in sirius is that cost pressure is not symmetric. Carriers pass through cost increases faster than they pass through decreases. This is not a model assumption but an empirically observed pattern across multiple crisis cycles. The sirius index scenarios reflect this asymmetry in recovery paths.

### Quality and auditability

The sirius index is built on three quality principles:

#### Source traceability

Every data point carries a provenance tag. When a reading changes, the source data that drove the change is identifiable and dated. There are no interpolated or estimated values without explicit labeling.

#### Versioning

Every weekly reading is permanently versioned. If a correction is needed (e.g., a source agency revises a data point), a new version is published with a changelog. Historical readings are never silently overwritten.

#### Self-correction

The sirius index includes a feedback mechanism that compares its forward projections against actual outcomes. When projections deviate from realized values, the model's parameters are adjusted. This creates a track record that can be audited: prediction accuracy, direction accuracy, and confidence band calibration are all measured and published.

sirius maintains an internal accuracy ledger comparing forward projections against realized market outcomes. We measure directional accuracy, confidence band calibration, and prediction timing, and publish track record summaries as part of the weekly brief.

### Limitations

The sirius index is a useful tool, not a perfect one. Users should be aware of the following limitations:

- **Spot market focus:** The sirius index measures cost pressure on spot and variable-rate freight. For long-term contracted rates with fixed surcharge clauses, actual exposure may be lower than the sirius reading suggests.
- **Public data dependency:** The sirius index relies on publicly available market data. If a source delays publication or changes methodology, affected readings are flagged but may carry lower confidence for that period.
- **Aggregation:** Each trade lane is a corridor average. Individual origin-destination pairs within a lane may experience different pressure depending on local conditions.
- **Crisis non-linearity:** In extreme events (full chokepoint closure, war escalation beyond precedent), sirius historical calibration may understate actual outcomes. Scenario bands attempt to capture this, but black swan events by definition exceed model assumptions.
- **No carrier-specific rates:** The sirius index measures market-level input pressure. Individual carrier surcharge formulas, update cycles, and commercial practices vary. The index does not model individual carrier behaviour.

### Comparison with other indices

The sirius index occupies a different position in the market intelligence landscape than existing freight indices:

| Dimension | Published rate benchmarks | sirius |
| --- | --- | --- |
| **What it measures** | Prices charged or paid to move cargo | Input cost pressure (upstream signals) |
| **Direction** | Backward-looking | Forward-looking (4-week forecast range) |
| **Decomposition** | A composite, a route-level rate, or a spread across contracts | 4 layers (fuel, capacity, surcharges, FX) |
| **Modalities** | Mostly ocean, some air | Ocean, air, road |
| **Currency layer** | Mostly USD | Trade-weighted FX per corridor |
| **Use case** | Benchmarking what the market charged | Anticipating where costs are heading |

These are complementary, not competitive. Published rate benchmarks tell you what the market charged last week, or what other shippers paid. The sirius index tells you where cost pressure is building next. Used together, they provide a complete picture: backward verification and forward anticipation.

#### See sirius in action

One reading a week, across all 10 trade lanes.

[See this week's reading →](https://tetrixx.ai/sirius#this-week)

## How to read the sirius index - and what it means for your freight budget

https://tetrixx.ai/sirius/methodology/pressure-model

Published 2026-05-27, updated 2026-09-26. By TetriXX AI & LSCMS.

One number. Four dimensions. Here is what it means and how to act on it.

A sirius reading of 118 means the cost pressure inputs on a lane are running 18% above the January 2026 baseline, which is not the same as a freight bill 18% higher. This article explains what that number means in practice, what drives it, and how to act on it.

4-layer contribution to composite reading - illustrative example · Layer contributions · ~120 · Composite reading · Fuel · Capacity · Surcharge · FX · Values are illustrative. Actual layer contributions vary by lane, modality, and market conditions.

### What a reading tells you

A sirius reading of 100 represents January 2026 - the baseline period. Every point above or below 100 is a percentage point deviation from that baseline. A reading of 118 means the cost pressure inputs are 18% above that level. A reading of 94 means they are 6% below it.

Four zones help orient the response:

EASING · Below baseline · Review locked-in rates. Renegotiation window is open.

STABLE · Near baseline · Normal variance range. Monitor weekly. No urgent action required.

PRESSURE · Elevated · Review contract renewal timing and modal alternatives. Act before the next invoice cycle.

CRITICAL · Significantly elevated · Activate contingency plans. Crisis-level pressure, typically from active chokepoint disruption.

sample reading · easing · critical

Zone thresholds are for orientation. Actual response depends on your freight cost structure and budget sensitivity.

What a reading of 118 does not mean is also worth stating: it does not mean your freight bill is 18% higher. sirius measures variable cost pressure on the inputs to freight pricing. Your actual freight invoice change depends on your contract structure, the proportion of your freight cost that is variable, and how carrier surcharge mechanisms pass through the pressure. Translating a reading into a cost delta on your own lanes and transport spend is a paid engagement, quoted on request.

### The four pressure dimensions

#### Fuel pressure

When crude oil moves significantly and carrier fuel surcharge schedules have not yet adjusted, fuel pressure leads the index. sirius has consistently flagged fuel pressure elevation ahead of carrier FSC announcements - the fuel layer registers the market move before it appears in quoted rates. A new reading is published every week, taken from Friday's market data.

Fuel pressure is the most responsive of the four dimensions to sudden market events. A geopolitical shock that disrupts oil supply typically surfaces in the fuel layer within 1-2 weeks. For shippers watching sirius, a fuel pressure spike is the earliest signal to check contract renewal timing and reassess modal mix assumptions.

#### Capacity pressure

Blank sailing programs reduce effective vessel capacity before rate announcements follow. sirius correlates booking lead-time signals and vessel scheduling data to detect capacity tightening before rates move. During chokepoint events, sirius captures the divergence between capacity and fuel pressure signals - the sequencing of these layers provides an early detection signal ahead of carrier rate announcements.

Capacity pressure typically builds and resolves more slowly than fuel pressure. Utilization spikes driven by blank sailing programs can persist well beyond the program itself - the duration varies by lane and carrier response. For procurement, a rising capacity dimension is the signal to act on contract renewal well ahead of the announcement window.

#### Surcharge pressure

BAF, EBS, and emergency surcharges compound on top of base rates. sirius tracks surcharge schedules across major carriers to model the real cost per TEU - not the quoted base rate. When a carrier declares a Peak Season Surcharge, the surcharge dimension registers the movement within the same week, before it appears in your invoices 30-90 days later.

Surcharge pressure is the most event-driven of the four dimensions. It can spike sharply when carriers declare emergency surcharges and resolve quickly if withdrawn. A rising surcharge dimension with stable capacity and fuel usually signals a tactical carrier pricing move rather than a structural market shift - useful context for contract negotiations.

### FX pressure

For shippers with USD-denominated contracts paying in EUR, MYR, or IDR, a 5% FX move can offset a 3% rate reduction entirely. sirius models the net impact per lane and currency pair. When USD strengthens significantly, FX pressure can contribute materially to the composite reading even if spot freight rates appear flat.

FX pressure tends to be a secondary amplifier rather than a primary driver in most market conditions. But it is consistently underestimated because it is invisible until the Currency Adjustment Factor resets on the next billing cycle. sirius surfaces the FX dimension weekly, giving finance teams the signal before the CAF adjustment arrives in AP.

**CAF vs FX pressure:** The FX dimension in sirius is not the same as the Currency Adjustment Factor on your carrier invoice. The CAF is a specific billing mechanism with its own update cycle. The FX dimension measures the underlying market pressure that eventually flows through the CAF - the two move in the same direction but with different timing.

### What sirius looked like during recent disruptions

When sirius detects elevated pressure on a trade lane, the reading tells you which dimension is driving it and how the pattern compares to historical precedents. The examples below are illustrative - they reflect the structural shape of how the four dimensions behaved, not precise reconstructed readings.

FE_EU lane · 142 · CRITICAL

Red Sea rerouting (Jan 2024)

FE_EU reached 142. Capacity pressure was the dominant driver - not fuel. Vessels rerouting via Cape of Good Hope added 14 days of transit time, compressing effective vessel availability on the lane before any fuel or surcharge movement registered.

(Illustrative example - not a forecast)

ME_FE lane · 131 · CRITICAL

Hormuz partial restriction (May 2026)

ME_FE reached 131. First detected as capacity pressure before fuel transmission registered. Chokepoint throughput reduction impacted vessel scheduling on the corridor within the first reporting week - fuel pressure followed as oil price uncertainty built over the subsequent two weeks.

(Illustrative example - not a forecast)

Multi-lane · 147 → 112 · 14 weeks

Post-COVID normalization (Q3 2022)

Index declined from 147 to 112 over 14 weeks - the asymmetric recovery pattern. Fuel pressure eased first, followed by capacity (as blank sailings were withdrawn), with surcharge pressure the last to normalize as carriers held emergency surcharges through the transition period.

(Illustrative example - not a forecast)

### The signal network: chokepoints and straits sirius monitors

Chokepoint throughput is a primary input to the capacity pressure layer. Four strategic straits account for a significant share of global maritime trade. sirius monitors vessel throughput at each chokepoint using vessel movement data from multiple sources, and applies a capacity adjustment when conditions deviate from historical norms.

Natural Earth · TopoJSON

Strait of Hormuz

The single most critical oil export channel, linking Gulf producers to global markets. Approximately 17 million barrels per day transit this strait in normal conditions.

sirius monitors this strait because a throughput reduction directly drives the capacity pressure layer on Middle East and Far East lanes, and amplifies the fuel pressure layer via oil price effects.

DEGRADED

Suez Canal

The primary shortcut between Asia and Europe, handling approximately 12% of global trade by value. Disruption forces vessels onto the Cape of Good Hope route, adding 7-14 days of transit time.

sirius monitors this canal because diversions add direct capacity pressure on Far East to Europe and Far East to North America lanes through vessel availability constraints and schedule unreliability.

NORMAL

Strait of Malacca

The primary gateway between the Indian Ocean and the Pacific, through which approximately 80,000 vessels transit annually, carrying roughly a third of global trade.

sirius monitors this strait because any throughput reduction creates immediate ripple effects on Intra-Asia, Far East, and Middle East to Far East lanes.

NORMAL

Panama Canal

The primary transit route between the Pacific and Atlantic, handling approximately 6% of global trade. Water level constraints have caused transit restrictions in prior drought years.

sirius monitors this canal because transit restrictions shift Far East to North America cargo to the Cape Horn route or to transpacific/rail combinations, affecting capacity pressure on multiple lanes simultaneously.

WATCH

Chokepoint status reflects the most recent sirius signal capture. Historical baselines from PortWatch/IMF and AIS vessel tracking data. Status indicators are illustrative of the monitoring framework - refer to the current sirius report for live readings.

#### See the current 4-layer breakdown

Fuel, capacity, surcharge and FX, weighted for this week.

[See this week's reading →](https://tetrixx.ai/sirius#this-week)

## Why freight cost pressure hits your books 30-90 days late

https://tetrixx.ai/sirius/methodology/invoice-lag

Published 2026-05-27, updated 2026-05-27. By TetriXX AI & LSCMS.

The signal moves. Your invoice follows. The gap between them is your decision window.

By the time your CFO sees a freight cost spike on the P&L, the market event that caused it is already 30-90 days old. sirius shows you what is coming - before it arrives in your invoices.

Invoice realization lag - illustrative example

T=0 · Spot spike · +30 days · M1 realizes · +60 days · M2 realizes · +90 days · M3 realizes

Market event (T=0) · M1 - Spot / index-linked (~30d) · M2 - Short-term contract (~60d) · M3 - Annual contract (~90d)

Numbers are illustrative. Actual lag varies by carrier, modality, and contract structure.

> **The insight:** By the time your CFO sees a freight cost spike on the P&L, the market event that caused it happened 30-90 days earlier. sirius closes that gap.

When a spot price shock reaches your invoices

Illustrative model - cost impact by contract type

[Chart: Stacked area chart showing invoice cost impact over 16 weeks for M1, M2, and M3 contract types after a spot price event at week 0]

M1 - spot / index-linked (~30d lag) · M2 - short-term contract (~60d lag) · M3 - annual contract (~90d lag)

Illustrative model only. Contract mix and transmission coefficients vary by modality and corridor. Modelling your own exposure, on your lanes and transport spend, is a paid engagement, quoted on request.

### The visibility gap

Supply chain cost management has a structural blind spot: market conditions and accounting reality are separated by weeks. When fuel prices spike or ocean capacity tightens, the effect on your freight invoices does not arrive immediately. It arrives weeks or months later, depending on how your freight is contracted.

This gap creates two compounding problems. First, finance teams are always reacting to the market of 30 to 90 days ago, not today's market. Second, procurement teams who watch spot market indices may believe costs are stabilizing while AP is about to receive a wave of invoices priced at the peak. These two teams can be looking at the same market and reaching opposite conclusions about their cost exposure - both correctly, because they are looking at different points in time.

Understanding the invoice lag is not a technical curiosity. It is the foundation of effective freight cost forecasting.

### How freight invoices work

A freight invoice is not generated at the moment a carrier quotes a rate. It is generated after the cargo moves, consolidated across multiple surcharge line items, and posted to AP when the carrier's billing cycle closes. Each of these steps adds time between the market event and the accounting entry.

The key surcharge components that change with market conditions are:

- **Bunker Adjustment Factor (BAF) / Fuel Surcharge (FSC):** Updated monthly or quarterly by most carriers, referencing a published fuel index from the prior period. A fuel price spike in week 1 typically reaches your BAF line in the invoice covering weeks 4 to 8.
- **Currency Adjustment Factor (CAF):** Updated on a similar cycle, referencing FX rates from the prior calculation window.
- **General Rate Increases (GRI) and Emergency Surcharges:** These tend to apply faster - sometimes with as little as 2 weeks notice on spot bookings - but take longer to flow through contracted rates.
- **War Risk / Piracy / Congestion Surcharges:** These can apply within days on spot freight but may be absorbed for a contract period on fixed-surcharge contracts.

The result is a layered stack of surcharge mechanisms, each on its own billing and update cycle, each referencing market data from a different point in the past.

### Three lag buckets

The sirius model organizes this complexity into three lag buckets - M1, M2, and M3 - representing the three dominant contract structures in global freight. Each bucket has a characteristic realization lag: the typical time between a market event and when that event is reflected in an invoice.

| M1 - Spot | M2 - Short-term | M3 - Annual |
| --- | --- | --- |
| ~30 day lag | ~60 day lag | ~90 day lag |
| Spot bookings, index-linked contracts, short-cycle tender agreements | 6-12 month contracts with quarterly surcharge reset mechanisms | Annual contracts with semi-annual or annual surcharge review clauses |
| Highest sensitivity to market moves. Cost pressure transmits fastest here. | Moderate buffering. Market shocks are absorbed for one quarter before resetting. | Strongest lag. Carrier absorbs cost pressure until the next review window. |

**On modal mix:** The M1/M2/M3 bucket breakdown interacts with your modal mix. Air freight is more heavily weighted toward M1 (spot-like update cycles). Ocean freight spans all three depending on carrier and contract type. Road freight often falls in M2. A company with 70% annual ocean contracts will see very different realization timing than one with 50% spot air.

### M1 - spot and index-linked

M1 represents freight purchased at or near spot market prices: open bookings, index-linked contracts that reprice monthly or faster, and short-cycle tender agreements where rates are reviewed at 30-day intervals. The typical invoice lag for M1 freight is approximately 30 days.

For a shipper with significant M1 exposure, a market event that occurs today will appear in invoices received roughly four to five weeks from now. The driver is not shipping time - it is billing cycle. Carriers accumulate charges across a freight period, issue an invoice at the end of the period, and the invoice enters AP workflow before it is posted and visible to finance.

M1 freight is the most responsive to market moves, which means it is also the highest-volatility contract tier. During a price spike, M1 costs can increase significantly within a single billing period. During a market correction, M1 is also the first to benefit.

### M2 - short-term contracts

M2 represents freight under short-term contracts, typically 6 to 12 months in duration, where surcharge mechanisms are reviewed quarterly or semi-annually. The contractual structure provides a buffer: the carrier absorbs short-duration market moves within a quarter and passes them through at the next surcharge reset. The typical invoice lag for M2 freight is approximately 60 days.

The 60-day figure reflects both the billing cycle delay (similar to M1) and the surcharge update cycle. A fuel price spike that occurs early in a quarter may not appear in the surcharge adjustment until the start of the next quarter, adding 30 to 60 days beyond the billing cycle lag.

M2 contracts offer shippers more predictability than spot freight, but they create a false sense of insulation. A company that has locked in M2 surcharge rates through the end of a quarter may see their spot-watching procurement team declare that freight costs are rising while their current invoices show no increase - yet. The reset is coming.

### M3 - annual contracts

M3 represents freight under long-term annual contracts where the surcharge mechanism is reviewed once or twice per year. The typical invoice lag for M3 freight is approximately 90 days, though in some structures the effective lag can extend to 120 to 150 days for shippers with mid-year contract anniversaries and semi-annual review clauses.

Annual contracts are the most insulating structure from a short-term volatility perspective - but they create the largest delayed exposure. A carrier absorbing a sustained cost increase across an entire contract period will reset aggressively at the next renewal. For shippers on annual contracts, the invoice surprise does not arrive as a small monthly increment but as a significant step change at renewal.

Annual contracts also create the longest gap between market intelligence and actionable response. By the time an M3 shipper sees the cost pressure in their invoices, the market event that caused it may have occurred three months ago. The window for contract renegotiation or modal rebalancing has typically already closed.

**Asymmetric transmission:** This asymmetric transmission behavior is well-documented in freight markets: carriers pass cost increases through faster than they pass decreases back. When calculating average lag, expect increases to land at the shorter end of the range and decreases to land at the longer end.

### Why CFOs see cost surprises

The invoice lag is the primary reason finance teams are surprised by freight cost variances. The organizational structure of most companies amplifies the problem. Procurement watches market indices and adjusts expectations based on current spot conditions. Finance books invoices as they arrive and compares them to the prior period. These two streams of information are telling different stories about the same underlying market - one current, one lagged - and there is typically no function bridging the gap.

The consequence is that budget variance meetings consistently feature freight as an unexplained line. "We saw this in the market three months ago, why wasn't it in the forecast?" The answer is usually that the forecast was built on current market data without modeling the realization lag of the actual contract portfolio.

A less obvious consequence is over-correction. When finance sees a large freight cost spike in Q2 invoices, they may assume the market is still at that level and budget conservatively for Q3 - even though the market has since corrected. The lag cuts both ways: it delays the pain but also delays the relief.

### How sirius models the lag

Rather than reporting only the current week's spot pressure, sirius calculates three parallel readings for each trade lane: the current market signal (M1 realization), the 60-day forward-realized reading (M2 projection), and the 90-day forward-realized reading (M3 projection).

This means a sirius user looking at the Far East to Europe lane today sees not just what the market is doing now, but what is already locked into their M2 invoices and what is building toward their M3 renewal window. Finance and procurement share the same temporal reference - they are looking at the same market through three different time horizons.

> Example - January 2026 (illustrative)

> FE_EU capacity pressure reached 118 on week 3. **M1 contracts** (spot and index-linked) repriced within 30 days. **M2 contracts** (6-12 month) repriced in week 7. **M3 annual contracts** absorbed the shock in the Q2 contract review. Subscribers had the forward warning 6 weeks before M3 repricing.

> (Illustrative example.)

**Practical use:** If you are building a quarterly freight cost budget, use the sirius M2 reading for your base case - this is what your short-term contracted invoices will reflect in 60 days. Use the M3 reading as your sensitivity case for annual contract renewals. The M1 reading is the signal to watch for intra-quarter variance from spot bookings.

#### See where the pressure is now

This week's index maps the gap between market movement and your books.

[See this week's reading →](https://tetrixx.ai/sirius#this-week)

## How spot, short-term, and long-term contracts realize cost pressure differently

https://tetrixx.ai/sirius/methodology/m1-m2-m3

Published 2026-05-27, updated 2026-05-27. By TetriXX AI & LSCMS.

How you buy your freight decides when a market move reaches your bill: about 30 days at spot, 60 on a 6 to 12 month contract, 90 on an annual one.

Pressure wave realization by contract tier - illustrative example

M1 · Realizes in ~30 days · M2 · Realizes in ~60 days · M3 · Realizes in ~90 days

A cost pressure event at T=0 propagates through each tier at its characteristic speed. Numbers are illustrative.

### Why contract structure matters

When a fuel price spike, a carrier surcharge increase, or a capacity squeeze occurs in the market, it does not reach every shipper at the same time. The speed of transmission depends almost entirely on how the freight is contracted: what tier the contract sits in, what surcharge update mechanism it contains, and what billing cycle the carrier uses.

Two shippers shipping identical cargo on identical lanes can see cost pressure arrive at entirely different times if they are on different contract structures. The shipper on spot booking will see an invoice reflecting the spike within 30 days. The shipper on an annual contract may not see it for 90 days - or until their next contract renewal.

This is not a flaw in the system. It is the intended design. Contract structure is a risk management tool: longer contract terms trade cost stability in the short run for delayed reset risk in the long run. The problem is that most finance and procurement functions treat contract structure as an administrative category rather than a temporal cost exposure parameter.

A shock at the feet travels up the spine. Here is how long it takes to reach your invoice. · Illustrative scenario - Brent +$30 shock

### The four pressure layers

Before explaining how each contract tier transmits pressure, it helps to understand the four layers the sirius model measures. Each layer has its own market dynamics and its own typical pass-through mechanism:

- **Fuel pressure:** Driven by bunker fuel, aviation jet fuel, and road diesel markets. Transmitted through BAF, FSC, EFS surcharges. Update cycles range from monthly (ocean) to weekly (air) to ongoing (road diesel spot).
- **Capacity pressure:** Driven by vessel utilization, container availability, and route-level blank sailing rates. Transmitted through GRI, capacity surcharges, and rate level. Slower to build, slower to release than fuel.
- **Surcharge pressure:** Driven by operational cost events - war risk, congestion, port delays, emergency operational surcharges. Can apply within days on spot freight but may be absorbed for the duration of a contracted period.
- **FX pressure:** Driven by currency movement against trade-lane-relevant pairs. Transmitted through CAF mechanisms. Update cycle varies by carrier - some monthly, some quarterly, some semi-annually.

The key insight is that these four layers are not synchronized. Fuel and FX update on relatively predictable billing cycles. Capacity pressure can be structural and sustained. Surcharge pressure can be sudden. A contract tier's transmission coefficient describes how much of each layer passes through and on what timeline.

### M1 - spot and index-linked contracts

M1 is the highest-transmission contract tier. Spot bookings, index-linked contracts, and agreements with monthly surcharge reviews all fall into M1. These arrangements have minimal buffering between the market event and the invoice line item.

For fuel pressure, M1 transmission is close to full pass-through within one billing period - typically 30 days. The carrier applies a fuel surcharge based on the published index from the prior period, so a fuel spike in week 1 appears in the week 4-5 invoice. For surcharge pressure (emergency surcharges, war risk), M1 contracts often apply within 1-2 weeks notice.

The M1 tier's strength is that it creates the most accurate real-time picture of market exposure. The weakness is that it also creates the most volatility: cost spikes in M1 land fast, without the smoothing that longer contract terms provide. For high-frequency shippers - those moving cargo weekly on the same lane - M1 exposure is manageable if the market is being watched. For sporadic shippers, it is the most unpredictable tier.

**Air freight tends toward M1:** Air cargo rates are closer to spot behavior than ocean. BAF and capacity adjustments in air freight are updated weekly or biweekly by most carriers, which means air freight cost pressure realizes faster than ocean freight on equivalent routes.

### M2 - short-term contracts

M2 represents 6-to-12-month contracts with quarterly surcharge reset mechanisms. These contracts provide a structural buffer: the carrier absorbs market moves within the quarter and reconciles them at the next reset point. The typical realization lag for M2 is approximately 60 days.

The 60-day figure reflects the combination of two delays: the billing cycle (15-30 days after cargo movement) and the surcharge update cycle (30-90 days depending on where in the contract quarter the event occurs). A fuel spike in month 1 of a quarter may not appear in the surcharge reset until month 4 of the following quarter.

For capacity pressure, M2 transmission is more complex. Carriers on short-term contracts may negotiate capacity lock-in, which means capacity pressure reduces availability but does not immediately translate into a higher rate. The pressure transmits at renegotiation, not at the market event.

For FX pressure, M2 contracts often use a quarterly CAF reset. A significant currency move mid-quarter may not appear until the following quarter's billing cycle. This is the most common source of unexpected FX variance in freight budgets: the rate moved three months ago, and the invoice is arriving now.

### M3 - annual contracts

M3 represents annual contracts with semi-annual or annual surcharge review clauses. These are the most common contract structure for large-volume ocean shippers - typically multinational manufacturers and retailers who negotiate directly with carriers or through freight forwarders. The typical realization lag is approximately 90 days, and in some structures can extend to 120-150 days.

M3 contracts offer the strongest short-term cost predictability. During a period of market stress, an M3 shipper may continue paying rates that are significantly below market for several months while M1 spot shippers absorb the spike immediately. This is the upside. The downside is reset risk.

When an M3 surcharge mechanism resets, it often resets on multiple dimensions simultaneously: fuel, FX, and capacity all reconcile at once. If the market has moved significantly over the contract period, the reset can be a significant step change rather than a gradual adjustment. This is the source of the "freight cost surprise" that appears in many Q4 or Q1 budget variances: an annual contract renewed in an elevated cost environment locks in the new baseline for the next 12 months.

**70% annual mix means lower volatility, not lower exposure:** A company with 70% of freight on annual contracts will show less quarter-to-quarter cost variance than a company with 70% spot exposure. But the total exposure over the contract cycle is similar - it is concentrated into fewer, larger reset events rather than spread across weekly invoice fluctuations.

### Transmission by pressure layer and contract tier

The following table summarizes how each pressure layer typically transmits through each contract tier. Sensitivity ratings are illustrative of typical carrier practice - individual contracts may differ.

| Pressure layer | M1 sensitivity | M2 sensitivity | M3 sensitivity |
| --- | --- | --- | --- |
| **Fuel** | High - monthly reset | Medium - quarterly reset | Low - semi-annual reset |
| **Capacity** | High - GRI applies quickly | Medium - buffered by volume lock | Low - absorbed until renewal |
| **Surcharge** | High - applies per booking | Medium - 2-4 week notice clause | Low - usually absorbed or capped |
| **FX** | High - monthly CAF | Medium - quarterly CAF | Low - semi-annual CAF |

Sensitivity ratings are illustrative. Actual transmission depends on carrier, contract language, and market conditions.

### How modal mix shifts the balance

The M1/M2/M3 split does not operate uniformly across modalities. Different transport modes have structurally different contract norms, which means the same shipper's ocean and air cost exposure can be in very different contract tiers simultaneously.

- **Ocean freight** spans all three tiers depending on volume and carrier relationship. High-volume shippers tend toward M3. Mid-volume shippers are typically M2. Spot ocean is M1.
- **Air freight** skews heavily toward M1. Airline cargo surcharges are updated weekly or biweekly. Even "contracted" air cargo rates are typically re-priced at 30-day intervals for most shippers outside the very largest consignors.
- **Road freight** is typically M2. Regional trucking contracts are usually quarterly or semi-annual. Spot road bookings can behave like M1, particularly in tight capacity markets.

A company with a 60% ocean / 30% air / 10% road modal mix will have a blended transmission coefficient that sits somewhere between M2 and M3 for its ocean component and closer to M1 for its air component. This is why two companies with similar spend can see very different quarterly cost variance patterns: the difference is often modal mix, not market exposure.

### How sirius weights contract mix

The sirius pressure model does not assume a universal contract structure. For each trade lane, the model applies a default modal and contract tier mix based on the structural characteristics of that lane.

The weekly report includes a full breakdown of the M1/M2/M3 weighting per lane and per modality. This allows a freight manager to map their own contract portfolio against the model's default assumptions and adjust the reading accordingly. A company with an above-average proportion of spot ocean booking on a lane would apply the M1 reading more heavily than the M3.

sirius calculates separate pressure readings for each contract tier and combines them into a composite reading using lane-specific weighting. The decomposition is surfaced in the weekly report so users can isolate the exposure most relevant to their purchasing structure.

This decomposition is the practical bridge between the market intelligence sirius provides and the budget variance conversations that finance teams need to have every quarter. Knowing that the Far East to Europe lane is at a composite reading of 118 is useful context. Knowing that your M3 component is lagged 90 days behind a peak reading of 124 is actionable.

**For budget planning:** When building a quarterly freight cost forecast, use the sirius M2 reading as your base case for the coming 60 days and the M3 reading as your sensitivity case for annual contract renewals. Run the M1 reading as your intra-quarter variance signal to track whether spot exposure will overshoot the base case.

#### See which tiers are absorbing pressure

M1, M2 and M3 exposure, lane by lane.

[See this week's reading →](https://tetrixx.ai/sirius#this-week)

## How sirius matches current market patterns against historical crisis analogs

https://tetrixx.ai/sirius/methodology/crisis-analog

Published 2026-05-27, updated 2026-10-12. By TetriXX AI & LSCMS.

When sirius detects a pattern that resembles a past crisis, it surfaces the match with a confidence score and the trajectory from the previous event. Here is what that output looks like in practice - and how the best-matching analog shapes the forward projection scenarios.

### The asymmetry of a crisis: fast shock, slow recovery

Crisis pressure builds faster than it resolves. A disruption that reaches peak intensity in 6 weeks may take 30 weeks to fully unwind. This asymmetry matters for how organizations budget and hedge: the spike is visible, the long tail is where cost exposure accumulates.

The chart below shows the structural shape of a typical crisis shock and how M1 and M2 contract types realize that pressure with a lag - M1 (spot and short-term, approximately 30-day invoice lag) and M2 (medium-term contracts, approximately 60-day lag).

Index pressure vs. M1 and M2 contract realization

Weeks from crisis onset (week 0 = confirmed disruption event)

Illustrative model. Index base = 100. Values reflect typical crisis shock and recovery patterns, not specific events. M1 = 30-day invoice lag; M2 = 60-day invoice lag. Peak amplitudes are compressed by lag: M1 ~90%, M2 ~75% of index peak.

### Crisis library coverage

The analog library covers publicly documented disruptions from 2020 to 2026. Each event was selected for signal diversity - fuel shocks, capacity events, chokepoint restrictions, and demand surges - and for resolution quality, meaning sufficient weekly data exists to encode the full trajectory.

COVID port congestion · COVID (cont.) · Suez blockage · Ukraine fuel shock · China lockdowns · Red Sea / Houthi · Panama drought · Red Sea (cont.) · Taiwan strait · ENSO / La Nina · US tariff shock · Peak season · Hormuz partial restriction · Peak season

2020 · 2021 · 2022 · 2023 · 2024 · 2025 · 2026

Fuel / capacity shock (amber) · Chokepoint disruption (red) · Environmental / regulatory (blue) · Geopolitical tension (purple) · Demand surge / seasonal (green)

Crises in the analog library are selected for signal diversity and resolution quality. All are publicly documented events. Bar height is proportional to event duration within each year column. Ongoing events shown through latest available data (May 2026).

Six-input pressure pattern match - illustrative example

[Figure: Radar chart showing current pattern vs best analog match across its six inputs: four pressure dimensions and two signal groups]

Current pattern (W21 - illustrative) · Best analog match (illustrative)

Values and shapes are illustrative. Actual scoring uses normalized pressure readings from the sirius model.

> Example output - Red Sea 2024 (illustrative)

> During the Red Sea escalation in early 2024, sirius identified a strong analog match against a historical capacity-led chokepoint disruption. The matched trajectory projected FE_EU pressure peaking and plateauing before gradual resolution. M2 contracts were flagged as the primary exposure. The forward signal was available to subscribers ahead of carrier rate announcements.

> (Illustrative example. Subscribers receive the current week's analog match in the weekly report, by Monday morning, Singapore time.)

### Why historical analogs

Freight markets are not random. They follow patterns that have recurred across crises: a fuel shock builds capacity pressure, capacity pressure drives surcharge increases, surcharge increases tighten the FX impact on cross-currency lanes, chokepoint degradation amplifies the whole sequence, and environmental cycles determine whether the network can absorb it. The sequence and relative intensity of these six inputs varies by crisis type, but the structural relationships between them are recognizable.

This means that when a new crisis begins to build in the data, it often resembles a historical precedent - not in its exact magnitude or geographic origin, but in its pattern across the four pressure dimensions and the two signal groups that feed them. Different crisis types produce distinct pressure signatures across them. The analog engine identifies which historical event most closely resembles the current configuration.

The crisis analog engine exploits this structural similarity. Rather than extrapolating from recent trend alone - which is unstable in the early stages of a crisis - it identifies which historical event most closely resembles the current pattern and uses the trajectory of that event as a calibration anchor for the forward projection.

### The crisis library

The analog library contains historical crisis events, each encoded as a time-series of multi-dimensional pressure readings normalized to the same base scale. The library covers a range of publicly documented disruptions across fuel shocks, capacity cycles, and chokepoint restrictions. A representative selection of events is highlighted in the weekly report.

Capacity-led · Port congestion, container shortage, and network saturation events

Chokepoint · Strait disruptions and route diversion events affecting transit capacity

Fuel-led · Oil price shock and carrier fuel surcharge cascade events

Geopolitical · Conflict-driven freight disruptions with compound capacity and fuel pressure

Additional · Further calibration events covering demand cycles, seasonal extremes, and mixed-driver shocks

Each crisis in the library is encoded as a weekly time series starting from the first week of measurable pressure elevation through to recovery, defined as all pressure dimensions returning to near-baseline levels. This produces a crisis "trajectory" with a shape, a duration, and a resolution pattern for each event.

**On crisis duration:** The library shows significant variation in crisis trajectory length. The 2021 Suez blockage resolved in days for the physical disruption but took 6-8 weeks for capacity pressure to normalize. The 2020-2021 COVID congestion took over 18 months to fully resolve. This variance in resolution timing is a key input to the analog engine's confidence band calculation.

The organism has responded before. Here is what it looked like - and what happened next. · weeks from crisis onset (week 0 = confirmed disruption) · sirius index · illustrative

### Six scoring inputs: four dimensions, two signal groups

Each crisis event - both historical and current - is represented as a six-dimensional vector at each week in its trajectory: the four pressure dimensions of the sirius pressure model (fuel, capacity, surcharge, FX), plus the two signal groups the engine reads alongside them (chokepoints and environment):

- **Fuel:** Normalized fuel pressure, capturing the combined effect of bunker, jet fuel, and road diesel market moves relevant to the lane being analyzed.
- **Capacity:** Normalized capacity pressure, capturing vessel utilization, container availability, and blank sailing data where available for the crisis period.
- **Surcharge:** Normalized surcharge pressure, capturing carrier-declared emergency surcharges, war risk premiums, and congestion charges as a proportion of base rate.
- **FX:** Normalized currency pressure, capturing the trade-weighted FX movement against the primary currency pair for the trade lane being scored.
- **Chokepoints & Straits:** Normalized throughput degradation at the key chokepoints relevant to the lane - Hormuz, Suez, Malacca, Panama. Derived from AIS transit counts relative to the 12-month rolling baseline. A partial restriction registers proportionally; a full closure registers at maximum.
- **Environment:** Normalized pressure from weather, seasonal cycles, and ENSO state. Captures typhoon season activation over Asia-Pacific lanes, monsoon cycle disruptions for Indian subcontinent routing, La Nina-driven drought impact on the Panama Canal, and the compounding effect of peak season demand on an already-stressed network.

Normalization maps each input to a 0-1 scale relative to the observed range across all historical crisis analogs. This prevents a crisis with extreme fuel pressure from dominating the similarity score simply because of scale differences.

### What an analog match looks like

Each week, sirius scores the current freight pressure pattern across these six inputs against every week in each historical crisis. The closest match becomes the primary analog. sirius then uses the trajectory of that historical event - how pressure evolved from the matched point forward - as the anchor for the forward projection.

The output delivered in the weekly report includes: the name of the primary analog event, the week-in-crisis it matched, the confidence score, and the projected path for the next 4 weeks based on how that event unfolded. Secondary analog matches are shown to indicate whether the forward trajectories converge or diverge.

**Illustrative example:** For the week of 19 May 2026, sirius identified a strong pattern match against Week 4 of the 2024 Red Sea escalation. The match was driven by a similar combination of capacity pressure leading fuel pressure - the same sequence seen in early-stage chokepoint disruptions. The analog trajectory projected ME_FE pressure peaking in weeks 3-5 before plateauing. Confidence was 78%. (Illustrative example - not a real sirius output.)

When the top-3 analog matches all point toward a similar trajectory, the confidence band in the forward projection is narrow. When they diverge - for instance if one analog resolved quickly and another persisted for 18 months - the band is wider, which is the correct reflection of genuine historical uncertainty.

### Reading the confidence score

The confidence score is published alongside each weekly sirius reading in the report. It tells you how strongly the current pattern resembles a historical precedent - and therefore how much weight to give the analog trajectory in your forward planning.

A high confidence score means the current pattern closely matches a historical crisis at a specific point in its trajectory. The forward projection from that analog is a well-anchored directional signal. A low confidence score means the current pattern does not closely resemble any historical precedent in the library - which is itself informative. It signals either an early-stage crisis still developing its signature, or a genuinely novel event.

When sirius shows low confidence, the appropriate response is to widen your scenario range and rely more heavily on the current-week reading than on the projected trajectory. A novel crisis with no close historical match should not be modeled using the average outcome of past events.

**What high confidence is not:** A confidence score of 85% does not mean there is an 85% probability the market follows the analog trajectory. It means the pattern is strongly similar to a historical event. The outcome still depends on how the current situation resolves - which may differ from the historical precedent for structural reasons the model does not observe.

### How analogs drive forward projections

When the analog engine finds a high-confidence match, the trajectory of the matched historical event becomes the anchor for the sirius forward projection on that lane. The trajectory is calibrated to the current situation's magnitude and extended forward from the matched point in the historical event.

sirius generates three forward scenarios - base, escalation, and de-escalation - grounded in the trajectories of the closest historical analogs. The scenarios reflect genuine historical variance in how comparable crisis patterns have resolved, not arbitrary percentage bands applied mechanically around the base case.

This structure means the scenarios are not arbitrary. They are grounded in observed historical outcomes on the closest precedents available. The uncertainty band reflects genuine historical variance in how similar crisis patterns have resolved, not a uniform percentage band applied mechanically around the base case.

### Why analogs are bounded and not extrapolated

The analog engine explicitly does not extrapolate beyond crisis resolution. Once the analog trajectory reaches its resolution window, the forward projection transitions to the sirius baseline model.

This is a deliberate design choice. The further a projection extends beyond the analog's resolution window, the less the historical precedent constrains it. A 16-week forward projection based on a crisis that resolved in 12 weeks is anchored for only the first 12 weeks. After that, any trajectory is equally plausible and the model would be manufacturing false precision.

The weekly report shows the analog's resolution window explicitly so users can see where the historically-anchored projection ends and where the baseline model takes over. The confidence band widens significantly at and beyond the resolution boundary.

**On novel crises:** When a new crisis type appears - one with no close historical analog - the engine will correctly report low confidence. This is the appropriate response. It means users should rely more heavily on the current-week reading and apply wider scenario bands than the model would generate for a well-matched historical precedent.

### Limitations

The analog engine is a useful calibration tool, not a prediction mechanism. Several limitations are important to hold in mind when interpreting analog-anchored projections:

- **The library is finite:** The library covers roughly 6 years of modern freight market dynamics across multiple crisis types. A genuinely novel event - a crisis type not well-represented in the library - will produce low-confidence matches. This is a signal to use wider judgment, not to ignore the reading.
- **Historical context differs:** Even a high-similarity pattern match does not guarantee that the future trajectory will follow the historical precedent. Market structure, carrier concentration, regulatory environment, and geopolitical context all change over time. The 2020 container shortage and a hypothetical 2027 container shortage with the same pressure pattern may resolve differently for structural reasons the model does not observe.
- **Six inputs are not the full picture:** The analog engine compares six inputs. Real crisis dynamics involve many more variables. The engine captures the dominant signals but may miss nuances in demand patterns, political resolution mechanisms, or infrastructure changes that would influence the trajectory.
- **Analog confidence is not probability:** A high confidence score means the current pattern closely resembles a historical precedent. It does not mean the outcome will follow that precedent with high probability. Both should be understood as directional guidance, not precise forecasts.

The sirius index publishes the primary analog name, its similarity score, and the confidence composite for every weekly reading in the report. This transparency allows users to assess the quality of the historical anchor for themselves rather than accepting the projection as a black-box output.

#### See the current pattern match

This week's index names the closest historical analog and its trajectory.

[See this week's reading →](https://tetrixx.ai/sirius#this-week)

# Blog

Source: https://tetrixx.ai/blog

- [Transport and Logistics Spend Management with AI](https://tetrixx.ai/blog/transport-logistics-spend-management-ai) (2026-10-05): Transport and logistics spend management with AI: why freight spend resists control, what a verified line-level record changes, seeing cost pressure before the invoice, and cost and carbon together.
- [Freight Billing Errors and How to Catch Them Before Payment](https://tetrixx.ai/blog/hidden-cost-freight-billing-errors) (2026-10-05): Freight billing errors explained: how often they occur and what they cost, the most common error types, why sampled audit misses them, and why catching them before payment changes the outcome.
- [Automated Freight Audit and AI Invoice Verification (2026)](https://tetrixx.ai/blog/automated-freight-audit-guide) (2026-10-05): Automated freight audit and AI invoice verification explained: rule-based versus AI-native audit, two-way and three-way matching, who decides what is disputed, and how to choose a platform.
- [Building a Freight Cost Outlook for 2026](https://tetrixx.ai/blog/building-a-freight-cost-outlook) (2026-09-14): A freight cost outlook is not a forecast. It is a structured view of the pressure signals that will drive your freight costs over the next 1 to 12 months. Here is how to build one that holds up in a board review.
- [Supply Chain Cost Intelligence, and What to Look For](https://tetrixx.ai/blog/supply-chain-cost-intelligence) (2026-09-14): Supply chain cost intelligence is the practice of monitoring freight cost pressure signals before they reach your invoices. This guide explains what it is, how it differs from freight market data, and what questions to ask when evaluating tools.
- [How Chokepoint Disruptions Flow Through to Your Freight Invoices](https://tetrixx.ai/blog/chokepoints-and-freight-costs) (2026-09-14): When a strait or canal faces restrictions, the cost impact does not arrive all at once. It travels through three waves of contract repricing. Here is how Hormuz, Suez, Malacca, and Panama disruptions reach your P&L.
- [Freight Rates and Cost Pressure, and Why Your CFO Needs Both](https://tetrixx.ai/blog/freight-cost-pressure-vs-rates) (2026-09-14): Freight rate indices tell you what the market charged last week. Cost pressure indices tell you what your invoices will look like in 60 days. Here is why the difference matters for P&L forecasting.
- [Why Your Freight Invoice Lags the Market by 30 to 90 Days](https://tetrixx.ai/blog/freight-invoice-lag) (2026-09-14): Freight costs move in the spot market today. Your invoice arrives weeks later. Here is how the M1, M2, and M3 invoice lag model explains the gap, and why it matters for your budget.
- [AI and Client Relationships in Freight Logistics](https://tetrixx.ai/blog/26-from-client-question-to-shipped-feature) (2026-08-11): AI is changing client relationships in freight, and not by removing people from them. What collapsed is everything after the conversation: what used to take a quarter now takes days.
- [Accidental Complexity in AI-Written Code (2026)](https://tetrixx.ai/blog/25-the-code-you-write-when-you-dont-know-the-system-yet) (2026-07-27): Accidental complexity is the second code path an AI agent builds when it does not know the system yet. Fred Brooks named the difference in 1986, and freight invoicing pays for it.
- [AI and Creative Thinking in Supply Chain Data Science](https://tetrixx.ai/blog/24-how-ai-enables-creative-thinking) (2026-07-26): AI collapses the cost of testing an unfamiliar method, so the constraint moves from execution to idea quality: the cross-field connections and honest judgment a machine still cannot supply.
- [AI Agent Token Costs and Model Choice in Logistics](https://tetrixx.ai/blog/23-not-every-package-needs-a-plane) (2026-06-26): AI tokens behave like freight: they cost money to move and there is only so much room. Optimizing an AI system turns out to be the same discipline as optimizing a supply chain.
- [MCP Server Infrastructure in Supply Chain AI (2025)](https://tetrixx.ai/blog/22-why-we-built-our-own-mcp-server-infrastructure-and-what-we-learned) (2025-12-17): MCP server infrastructure for supply chain AI agents: why a single monolithic agent fails on logistics data, how Agent Forge wires specialized agents to MCP servers, and three lessons from building it.
- [AI for Logistics Data in Freight Invoice Auditing](https://tetrixx.ai/blog/21-unlocking-the-value-of-logistics-data-with-ai) (2025-12-15): AI for logistics data turns fragmented freight rates, surcharges and invoices into a harmonized foundation for cost control, supplier benchmarking and proactive risk management.
- [Supply Chain AI Ground Truth and Data Foundations](https://tetrixx.ai/blog/20-is-your-supply-chain-ai-flying-blind-the-critical-need-for-ground-truth) (2025-12-15): Supply chain AI is only as reliable as its ground truth. How backend logic, APIs, data strategy and governance turn messy freight data into inputs an AI model can safely decide on.
- [AI Agents for Supply Chain Data Queries (2025)](https://tetrixx.ai/blog/19-beyond-manual-reports-why-your-supply-chain-data-needs-a-conversation-not-just-a-chart) (2025-08-04): AI agents let supply chain teams ask their data questions in plain English instead of building manual reports: what an agent is, an RFP generated end to end, and the data quality it needs.
- [Product Marketing Internship at TetriXX AI, Singapore (2025)](https://tetrixx.ai/blog/building-vision-into-reality-a-product-marketing-summer-at-tetrixx) (2025-07-30): A product marketing internship at TetriXX AI in Singapore: onboarding on 300 internal documents, the Paris product launch, a full website redesign, and the first external newsletter.
- [Legacy ERP Systems and AI Innovation in Supply Chain](https://tetrixx.ai/blog/legacy-erp-systems-ai-innovation) (2025-04-04): Legacy ERP systems promise integration but often block it. On data silos, the rigidity-versus-agility paradox, and pairing ERP with specialised AI agents.
- [Supply Chain Data Quality and Invoice Overbilling](https://tetrixx.ai/blog/freightproblemsonesolution) (2025-03-25): Poor data quality costs supply chains up to 6% of audited freight spend in overbilling. How freya, TetriXX AI's agent, tackles audit inefficiency, compliance risk and delayed reporting.
- [AI ROI and Data Quality in Supply Chain Management](https://tetrixx.ai/blog/ai-roi-data-quality-supply-chain) (2025-02-21): AI in supply chain management pays back only on clean data: how AI lifts ROI, why silos and inconsistent formats undermine it, and the governance and pilot approach that balances the two.
- [AI in Supply Chain Performance and Bias Mitigation](https://tetrixx.ai/blog/ai_in_supply_chain_boosting_performance_mitigating_bias) (2025-02-13): AI sharpens forecasting and automates procurement, but it can also encode bias. How to spot biased algorithms and the strategies that mitigate them.
- [AI Expectations in Supply Chain Management (2025)](https://tetrixx.ai/blog/ai_in_supply_chain_expectations) (2025-01-31): AI in supply chain management: predictive analytics that cut waste, automation that lifts warehouse and route efficiency, and what Amazon, Walmart and UPS actually run today.
- [Artificial Intelligence Types and Core Technologies](https://tetrixx.ai/blog/ai101) (2025-01-21): Artificial intelligence for beginners: narrow, general and super AI, and how machine learning, deep learning, NLP, computer vision, large language models and RAG each fit together.
- [AI Hallucination in Freight Invoice Auditing](https://tetrixx.ai/blog/navigating_ai_hallucination_in_the_freight_industry) (2025-01-07): AI hallucination in freight invoice auditing: why LLMs invent numbers on unstandardised freight data, and how fine-tuned models, algorithmic rules and human review keep audit outputs reliable.
- [Manual Invoice Errors in Maritime Freight Auditing](https://tetrixx.ai/blog/genz_outlook) (2024-10-25): Maritime freight still runs on hand-filled spreadsheets and paper invoices. How an 18-cent error per invoice compounds into a 3.6% hit on freight spend.
- [Last Mile Delivery Challenges and Technology Solutions](https://tetrixx.ai/blog/one_last_push) (2024-03-19): The Last Mile decides whether supply chain succeeds: cost, emissions, infrastructure and area density - and the analytics and optimisation tools that help.
- [End-to-End Traceability Technologies in Supply Chain](https://tetrixx.ai/blog/end-to-end-traceability-technologies) (2024-03-11): End-to-end traceability in supply chain runs on six technologies: an enterprise-wide data model, real-time monitoring, graph connectivity, blockchain, advanced analytics and cloud computing.
- [End-to-End Traceability in Supply Chain, Part 1 (2024)](https://tetrixx.ai/blog/where_is_my_thingpart_1) (2024-03-04): End-to-end traceability in supply chain: what it covers - item of interest, start of journey, point of use, point of transformation, end of journey - and the four reasons it fails.
- [Empty Container Repositioning in Supply Chain](https://tetrixx.ai/blog/empty-container-repositioning) (2024-02-26): Empty container repositioning is one of the costliest problems in supply chain logistics. The data, coordination and trade imbalance challenges behind it, what it costs, and why companies fail at it.
- [Supply Chain Disruption Types and Resilience Strategies](https://tetrixx.ai/blog/supply-chain-disruption-types-resilience) (2024-02-05): Supply chain disruptions run from natural disasters and geopolitical tensions to pandemics, cyberattacks and failing suppliers. Five disruption types, and seven strategies that build resilience.
- [Digital Transformation in Supply Chain Management](https://tetrixx.ai/blog/digital_transformation_in_supply_chain) (2024-01-21): Digital transformation in supply chain management: the three reasons enterprises take it on, the seven pillars a programme rests on, why efforts fail, and what makes them succeed, from TetriXX AI.
- [Supply Chain Data Strategy Elements and Benefits](https://tetrixx.ai/blog/elements_of_a_successful_data_strategy) (2024-01-20): Supply chain data strategy rests on six elements: governance, data collection and integration, predictive modelling, security and privacy, scalable cloud infrastructure, and continuous improvement.
- [Real Decision Time in Supply Chain (2024)](https://tetrixx.ai/blog/2-real-decision-time-in-supply-chain) (2024-01-18): Supply chain agility is not how much stock you hold, it is how fast you can re-plan, re-route and re-price. The four delays that make up a decision cycle, and how to measure yours.
- [Excel in Supply Chain Management, Pros and Cons](https://tetrixx.ai/blog/excel-or-excel-mentality) (2024-01-17): Excel earned its place in supply chain management, but version control and scale are where it breaks. The mentality is harder to shift than the tool.