Operanta AI Suite · KYTPublic Q3 2026

Know Your Transaction — explainable, and defensible in audit.

KYT vets every crypto transaction and returns a verdict you can defend, not a black-box score — a signed Evidence Pack on every verdict, and a deep trace that follows the money when a case needs it.

See how it works ↓
FATF R.15EU TFRMiCAVARADFSA · FSRASAMA
Powered byOperanta Agent Security Platform

Our flagship Security & Evidence layer for AI — so the AI doing the vetting is itself secured and verifiable.

The stakes

Your KYT tool gives you a score. Your regulator asks for a reason. That gap is your liability.

Under AML/CTF supervision, what counts as compliance has changed. A policy document is no longer enough — the regulator asks for operational evidence: proof of what the system actually did on a given transaction, and why this transfer was cleared or that wallet was blocked. A control you can't demonstrate counts as a control you don't have.

The decision to clear or block a transaction is a regulatory decision. If you can't explain it, you shouldn't have made it.

What you actually get

A verdict you can defend — and a trace that follows the money.

Two things no black-box KYT gives you together.

A · The verdict

A verdict you can defend

  • ExplainedHuman-readable reasoning your MLRO can put straight into a SAR.
  • Fund-flow checkedJudged on the actual movement of funds, not a wallet’s reputation — far less noise.
  • SignedSealed in a tamper-evident Evidence Pack, verifiable offline.
B · The trace

A trace that follows the money

  • Deep tracePast where clustering tools stop, through a mixer to the terminal wallets.
  • DemixReconstructs which funds went where, and where the money settled.
  • Ranked by confidenceEvery hypothesis tiered by likelihood — a lead, never a fact.
How it works

Every transaction, three independent checks, a signed pack — automatically, at volume.

Each transaction runs three parallel, explainable checks, reconciled by deterministic policy into a single verdict. Clear cases are auto-decided; borderline cases route to an analyst with the context already assembled. Every outcome is sealed as a signed Evidence Pack.

01

Sanctions & watchlist screening

OFAC, OpenSanctions, UN Security Council, regulator-specific and your own lists — every match with its full rationale.

02

Counterparty exposure

Multi-hop tracing of actual transfers, cluster detection, exposure across chains of accounts.

03

Suspicious-pattern detection

Behavioural models on structurally suspicious patterns, deeper than rule-based detectors.

The artefact

The artefact the regulator actually wants.

The Evidence Pack is the record of everything that happened to one transaction across the whole pipeline — the verdict, the independent check results, the policy rules that fired, the reasoning, and a cryptographic signature.

When the regulator asks why a transaction was cleared or blocked six months ago, the answer is one download away.

Investigate / Trace

When funds hit a mixer, the standard tool stops. Operanta keeps going — to the terminal wallets.

Screening tells you whether an object is clean. Investigate / Trace tells you what happened to the funds— where they came from, where they went, and how they were laundered. Same engine as screening; it goes where the market won't.

Deep trace — up to ~100 hops

The whole history of how funds moved, not a short horizon.

Ranked hypotheses

The path is tiered weak / medium / strong — prioritise by likelihood, not blindly.

Demix — the key capability

Reconstructs which of the subject’s funds went where, and highlights the terminal wallets where the money settled.

The investigation becomes evidence

The result attaches to the signed chain (Export to Case / Export to Evidence) — part of the defensible record.

See it

See it on a transaction.

The verdict, the proof, the trace, the queue — what your team actually works in.

Sanctions, counterparty, patterns, policy rules fired, plain-language reasoning, and the route to manual review — signed.

The foundation

You can't vet transactions with an AI that isn't secure itself.

AI isn't safe by default — a model can be steered by on-chain prompt-injection, or quietly swapped between audits. KYT runs on Operanta Agent Security Platform, so the AI is bounded (an architectural bound, not a probabilistic filter), its evidence is signed, and the whole thing maps to EU AI Act Article 15(5). The same foundation runs and secures your own AI agents too — not just our KYT.

Who it's for

Built for crypto operators under AML/CTF supervision — at any size.

Exchanges and OTC desks · custody and wallet services · payment processors and EMIs with crypto rails · DeFi protocols and stablecoin issuers.

The trace challenge

Give us an address where your current vendor lost the trail at a mixer — we reconstruct the route to the terminal wallets and show the hypotheses with honest confidence levels.

Then run Operanta in shadow mode beside your current KYT and compare on your own traffic — before you switch anything. Public from Q3 2026.

Or email us: hello@operanta.ai