MTekLabsSee supply chain risk build up.
Before it crosses the line.
Tributary connects what is moving, where it is going and who is behind it. It catches the risk no single shipment shows, ranks where to look next, and cites the evidence for every finding. A person decides what happens next.
For regulated shippers, procurement and trade compliance teams, and the agencies that oversee them.
-
Sources readShipment recordsOwnership linksSanctions lists
-
Deterministic indicatorReceipts building up across three sister companiesEach site under its threshold, together over · rule cited
-
Human decisionReviewer opens a caseNamed reviewer · time · reason recorded
-
RecordAdded to the audit trailHash-chained, tamper-evident
Supply chain risk
hides in the aggregate.
Each shipment can look clean on its own. The risk shows up across them: in what builds up at a site, in routes that change mid-journey, and in who owns whom.
Split across affiliates
Receipts are spread over sister companies, so no single site ever crosses a threshold.
Changed mid-route
A destination changes, a tracking signal goes dark, cargo moves ship to ship. The paperwork still looks clean.
Hidden owners
The consignee is new, but its parent is on a sanctions list. Screening the name alone misses it.
Scattered evidence
Manifests, carrier scans, vessel positions, registries and lists sit in different systems, and most tools judge one shipment at a time.
The pattern only exists in aggregate, and it's only useful before the goods move on.
Four questions.
One defensible picture.
Tributary brings together your shipment records, carrier and vessel tracking, ownership data and official lists, and answers the questions a reviewer, an auditor or a court will ask.
What is it?
Identifies each material from its CAS number, UN number, name or synonym, and shows every regulated list it's on, with the citation. A policy gate returns clear, review or block.
Where is it going?
Follows shipments by sea, rail and road through carrier, customs and vessel tracking, and flags route changes the moment they happen.
Who is behind it?
Maps shippers, consignees, carriers and owners into one graph, resolves who ultimately controls each site, and screens them against official sanctions lists.
What is building up?
Tracks what each site, and each group of related companies, receives over time, flags it before a limit is crossed, and ranks where to look next.
Start where the stakes
are highest.
Tributary goes deepest where a missed pattern carries legal, safety or security consequences. The same engine extends to new regimes as new reference data, not a rebuild.
Chemical & hazmat compliance
Check every shipment against DOT/PHMSA, OSHA, EPA, DEA and CFATS obligations, catch accumulation at your own sites, and work each finding to a recorded decision.
Supplier & material screening
Screen suppliers and their owners against sanctions lists before you buy, and gate restricted materials by your policy: clear, review or block.
Border & precursor intelligence
Find precursor networks before shipments reach a port of entry: hotspots, build-up across related companies and route changes on one operations picture.
Maritime sanctions & trade compliance
Tracking gaps, ship-to-ship transfers, loitering and destination changes, combined with ownership-aware screening. Runs on vessel and trade data you bring or we license.
Export controls & other restricted goods
Rules, material lists and screening lists are versioned, cited data. A new regime starts as new reference data and rules, scoped and reviewed with you.
Works with MatryxAI. MatryxAI helps you choose a supplier. Tributary watches what happens after: what moves, where it goes and who is behind it. MatryxAI already calls Tributary for its material checks and supplier screening.
Explore MatryxAIFrom the first record
to a decision you can defend.
One continuous loop. Each step leaves an entry in a per-tenant, hash-chained ledger that can be verified at any time.
Ingest
Shipment records, carrier and customs scans, vessel positions, ownership, lists, news
Detect
Cited rules, accumulation, route changes, screening
Rank
Hotspots scored on early-warning signals, as of the data
Review
Acknowledge, dismiss with a reason, or open a case
Hand off
Cases that can't close without their evidence, exports, signed webhooks, Throughline
Every shipment checked. Every finding cited.
Deterministic rules check shipping papers, placarding, packing groups, safety data sheets, training records, storage permits and regulated-transaction records against DOT/PHMSA, OSHA, EPA, DEA and CFATS obligations.
Each finding names the regulation it tests, carries a rule id and version, and is worked to a recorded decision in the exception queue. The same records always produce the same findings.
Caught on the way up.
Tributary tracks what each receiving site takes in over a rolling window, and flags the moment a material, or a combination of materials, approaches a reporting, storage, segregation or inventory obligation. Each flag cites the rule it was tested against.
When a new material needs watching, one person proposes it with a watch threshold and a different person approves it. Consignments held earlier are then re-evaluated, and every flag cites the watch threshold instead of pretending it is a regulatory one.
The split no single site shows.
Tributary resolves each receiving site to its ultimate controlling party through ownership links above a control threshold, then runs the same detector over the group as if it were one consignee.
It raises a group-level indicator only when no member site raises it alone, so reviewers see the new pattern without a pile of duplicates. Each indicator shows every member's contribution and the obligation it was tested against.
Hotspots, ranked before anything crosses the line.
Sites and groups of related companies are scored on early-warning signals: approaching or over a threshold (with a straight-line projection from the recent receipt rate), accelerating receipts, a new consignee, related materials arriving together, inbound route changes, and build-up that only shows at group level. Each signal comes with its evidence.
Rankings are computed as of the end of the data, not the clock, so the same data always gives the same ranking. Anyone with the right role can Watch a hotspot: the server builds a private daily alert from what was shown, so nobody needs rule-editing rights.
Route changes, as they happen.
Eight versioned detectors watch voyages and truck legs: undeclared port calls, destination changes, tracking (AIS) gaps, ship-to-ship transfers, loitering, route diversions, road detours and corridor tripwires. Severity rises automatically when the cargo is on a regulated list.
Analysts draw corridors and areas of interest with enter, exit and dwell tripwires, and see port congestion and a movement dossier for every receiving site. Re-running the sweep never creates a duplicate alert.
Who is really behind the consignee.
Shippers, consignees, carriers and owners form one ownership graph that analysts can expand a hop at a time. Screening runs against the OFAC SDN list, the UN consolidated list and the US Consolidated Screening List, and applies the OFAC 50% rule through ownership links.
Every candidate match carries a score that explains itself: which name, identifier or feature agreed. A match is never a conclusion. Each disposition is a reviewer's decision with a required note, and re-importing a list marks dropped entries as delisted instead of erasing the history behind existing matches.
The panels above are simplified representations of live product surfaces, populated from Tributary's synthetic demonstration scenarios. Every company, person, vessel and list entry in them is fictional and labelled as such; real places appear only as locations. Nothing on this page describes a real investigation or a customer outcome.
Proactive,
without guessing.
Early warning is only worth acting on if a reviewer can explain it, a supervisor can replay it and a court can follow it. Four properties make that true, whether or not anyone is watching.
Indicators, not verdicts
Every output carries its severity, confidence, the rule and version that raised it, declared versus observed values, and the evidence. Acknowledging, dismissing and opening a case are explicit, audited human decisions.
Reproducible as of the data
Detection is pure, versioned code. Hotspots are ranked as of the end of the data, not the clock, so the same data gives the same ranking whenever you run it.
Human decisions stay decided
A recomputation never re-opens an indicator a person acknowledged, dismissed or escalated. An unreviewed indicator whose evidence has gone is retired, and comes back if the evidence does.
Absent is never clear
Every screening result states which lists and versions were checked. A screen with no real list behind it says nothing was screened, instead of coming back clean.
Where Tributary sits against the alternatives.
Shipment-level compliance tools answer “is this consignment in order?” Model-scored risk platforms answer “how risky does this look?” Tributary answers the question in between: what is building up, where, and why, with an answer you can replay.
| Capability | Shipment-level compliance tools | Model-scored risk platforms | Tributary |
|---|---|---|---|
| Accumulation across shipmentsBuild-up at a receiving site over a window | No | Varies | Yes, cited per obligation |
| Splitting across affiliatesBuild-up across commonly owned sites | No | Varies | Yes, group-only indicators |
| Early-warning rankingWhich sites to watch next | No | Model score | Explained signals, reproducible as of the data |
| Route-change indicatorsGaps, transfers, diversions, detours | No | Often | Eight versioned detectors |
| Why it firedEvidence and rule behind each output | Rule reference | Limited | Evidence, rule version and citation on every indicator |
| Human decisions preservedReviews survive a recomputation | n/a | Varies | Never re-opened by the system |
| Tamper-evident recordA verifiable chain of every decision | Audit log | Varies | Per-tenant hash-chained ledger |
| Agent accessUse from your analysts' own AI tools | No | Varies | Read-only MCP tools |
One core.
Two editions.
The same engines serve shippers and the agencies that oversee them. A request can do only what both its token and the tenant's edition allow, for people, services and AI agents alike.
For the regulated shipper
Distributors, manufacturers, 3PLs and procurement teams.
Your own shipment records plus public references. Deterministic, cited findings, with no behavioral scoring.
- Cited compliance rules across DOT/PHMSA, OSHA, EPA, DEA and CFATS obligations
- Site accumulation and material thresholds
- Supplier, customer and party screening, with list coverage stated
- Audit dossiers, scheduled exports and cases that can't close without their evidence
For the analyst and the officer
Customs, border, counter-narcotics, counter-proliferation and trade compliance teams.
Adds licensed vessel and trade data to your own, and confidence-scored indicators about third parties.
- Route-change indicators, corridors and tripwires
- Build-up across related companies, and hotspot ranking
- Ownership graph and ownership-aware screening
- Second-person approval before any hand-off, escalation or disclosure export
Pressure is moving upstream.
From the single shipment to the network behind it. These public figures describe the environment Tributary is built for. None of them is a result of using Tributary.
The quantity threshold for propionyl chloride, a DEA List I chemical, from 9 Apr 2026.
Thresholds are falling. When every receipt is regulated, accumulation has to be tracked across all of them, not just the large orders.
Source: Federal Register
Cartels and gangs designated as Foreign Terrorist Organizations by the US State Department in Feb 2025.
Who receives your goods, and who owns them, now carries terrorism-related stakes for suppliers, shippers and agencies alike.
Source: Federal Register
Additional funding requested in the DHS FY2027 budget for the National Targeting Center's anti-drug campaign.
Agencies are funding targeting that finds networks before shipments arrive, not after.
Source: CT Strategies
Fentanyl seized by US Customs and Border Protection in FY2025, the lowest total since 2021.
Seizures catch what has already been made and shipped. The earlier signal is upstream, in the supply chain.
Source: CBP via WOLA
Worth being direct about.
The fastest way to waste an evaluation is to find out in week six that the scope was never right. Here is where Tributary earns its place, and where it doesn't.
A strong fit when
- You need to see goods accumulating across many shipments, not judge them one at a time
- The parties you deal with, or watch, can split receipts across affiliates, fronts or new consignees
- Your findings may end in an audit, an enforcement action or a dispute, and must be reproduced exactly
- You want early warning you can act on today, with the evidence attached
- Every escalation, disclosure and hand-off must be a documented human decision
- Your analysts already use AI assistants and need a governed way to query the picture
Not the right tool when
- You need demand planning, ETA prediction or inventory optimization. Tributary covers risk and compliance in what moves, not planning
- You want supplier financial-health or ESG ratings. Tributary reports evidence on materials, movement and ownership, not model scores
- You want a system that decides who is at fault. Tributary raises indicators; people make determinations, by design
- You need personal-data OSINT, CCTV feeds or cyber reconnaissance. These are excluded by charter
- You need long-running investigative case management. Tributary hands off to Throughline for that
- You need live global vessel tracking on day one without a data license. Intelligence features run on AIS and trade data licensed by you or by us
Ask for the artifacts
before the demo.
Demos are easy to stage. These are harder, which is what makes them worth asking for, of us and of anyone else you are evaluating.
The same ranking, run on different days
Load the same data window, rank the hotspots today and again tomorrow, and compare the two. Identical, or the claim is false.
A dismissal that stays dismissed
Dismiss an indicator with a reason, re-run the sweep, and watch it stay dismissed. Remove the evidence behind an unreviewed one, and watch the system retire it, then re-open it when the evidence returns.
Audit-chain verification
Every state change a regulator would ask about is written to a per-tenant hash-chained ledger. Run the verification, alter a record, and run it again.
The role boundary, enforced by the server
An officer can Watch a hotspot but can't write alert rules. The person who proposes a new tracked material can't approve it. Our smoke test signs in as every role and proves the server refuses what each one must not do.
A screening result that admits its coverage
Screen a party with only demonstration lists loaded. The result names the lists and versions checked, and says first that this is not a real screening.
The ones that actually
get asked.
Is Tributary only for precursor chemicals?
No. Precursors are where it started and its deepest use case. The engine tracks shipments, routes, parties and any material on its reference lists. Today those lists cover chemicals and hazardous materials under DEA, DHS CFATS, EPA and DOT rules, plus the OFAC, UN and US consolidated sanctions lists. A new regime starts as new reference data and rules, not a new engine.
How is Tributary different from MatryxAI?
MatryxAI helps you choose a supplier and defend the award. Tributary watches what happens after: what moves, where it goes, who is behind it and what builds up. MatryxAI already calls Tributary for its material checks and supplier screening.
Does Tributary decide who is at fault?
No. It raises indicators with their evidence, confidence and rule. Reviewers and officers acknowledge, dismiss with a reason, or open a case, and each of those is an audited human action. No rule infers intent, and no automated process acts on an indicator.
Is there a language model in the scoring path?
No. Detection, accumulation and hotspot ranking are deterministic, versioned code, so the same inputs produce the same outputs. Where we use AI, such as reading a material from messy manifest text or drafting a cited briefing, it sits outside the path that produces indicators and scores, behind guardrails and evaluation gates, and a person accepts or discards every draft.
What data does it need?
The Compliance edition runs on your own shipment records plus public references: the regulations behind each rule and the official sanctions lists. Carrier and customs tracking connect with your own credentials. The Intelligence edition adds licensed vessel-position and trade data, which you can bring or we can license for the lanes you care about.
How do we start tracking a new material?
One person proposes the material with a watch threshold and a reason. A different administrator approves it, and the threshold can't change while the proposal is pending. On approval, consignments held earlier are re-resolved and evaluated, and every resulting flag cites the watch threshold set at approval.
Can our analysts use it from their own AI assistants?
Yes, through six read-only MCP tools covering indicators, routes, screening, entity expansion, corridor activity and hotspots: list_indicators, shipment_route, screen_party, expand_entity, corridor_activity and get_hotspots. Agents can ask questions; they cannot change anything. Every change to a case or a disposition goes through a person.
What about personal data and open-source intelligence?
Open-source signals come from news event data, de-duplicated and linked to sites and materials with their provenance shown. Tributary does not scrape personal data, use CCTV feeds, run reconnaissance or query breach data, and it never will.
Where does it run?
Tributary ships as containers with PostgreSQL and Temporal, so it can run as a dedicated instance in your own cloud account. Identity comes from your OpenID Connect provider, every query is scoped to your tenant, and outbound webhooks are signed and guarded against server-side request forgery. Optional AI features run on Amazon Bedrock or Google Vertex AI.
Bring us the lane
you worry about most.
Not the clean one. The supplier group that never quite crosses a threshold, the route that keeps changing, the consignee nobody can place.
The picture, with its working shown.
- 1Your hardest laneA supplier group, a corridor or a set of consignees you want to understand.
- 2The picture assembledMaterials, movement, parties and accumulation together, on our synthetic scenario or a sample of your own records.
- 3Every finding explainedEach indicator with its rule, version and evidence. You decide what is worth pursuing.