⚠️ Portfolio / skills showcase. Entropa has zero real customers — this entire site and system exists to demonstrate real, working engineering, not to sell a live product. Pricing shown below is free / demo / test only.

Entropa_

Your examiners don't take your word for it. Now you don't have to ask them to.

Entropa gives banks and fintechs a signed, independently verifiable record of every AI-driven fraud, KYC, and AML screening decision — provable to an examiner, auditor, or disputing customer without trusting your own database, or Entropa's servers, or anyone's word for it.

Free · demo / test only · live in minutes

Getting started

Watch how easy this actually is — no sales call, no manual approval, live in minutes.

1. Sign up Pay by card, ~1 minute 2. Log in Emailed code, no password 3. Connect your agent API key, one REST call 4. Submit & verify Signed receipt, no trust required 5. Manage anytime Keys, audit trail, dashboard
Try the demo →

The problem

AI agents now clear transactions, screen customers, and flag suspicious activity — decisions your BSA/AML examiners can ask you to justify, sometimes years later. When they do, "trust our logs" isn't proof. A database row can be edited after the fact, and any examiner evaluating your program knows it. Recordkeeping requirements under the Bank Secrecy Act and FinCEN's own rules don't bend for "the AI decided it" — you still need a record that holds up.

We never see your customer data

You hash the decision yourself, on your own systems, before anything reaches Entropa. We only ever receive that hash — never the underlying customer name, account number, transaction detail, or decision content. That's not a policy promise; it's structural. There's nothing sensitive here for a vendor-risk review to find, because there's nothing sensitive here at all.

For a bank running third-party risk management under guidance like OCC Bulletin 2013-29, that matters directly: a vendor that structurally cannot access sensitive data is a fundamentally lighter due-diligence lift than one that stores it and promises to protect it.

How it works

Entropa logs every AI-agent action you send it, so nothing depends on trusting your own database. Here's the whole path, end to end:

1
Your agent acts
It decides something, approves something, does something real — any action you need governed.
2
You hash it
Compute a hash of the action. The raw data never leaves your system.
3
Submit the hash
One plain REST call — any language, no SDK. You get back 202 Accepted and a tracking ID immediately, so your own system is never stuck waiting on us.
4
It's appended to the log
Your record becomes a leaf in a Merkle transparency log — cryptographically hashed and chained.
5
A checkpoint is signed
The log's current root is signed into a checkpoint — the one thing every later proof gets checked against.
6
You fetch the receipt
Use your tracking ID to pull the receipt — usually ready within moments. Each record page runs and displays a real inclusion-proof and signature check, not just a claim.
Try the demo →

Make your own audit trail readable

Entropa only ever sees a hash — that's the whole privacy guarantee. But you can attach a short, optional label your own agent generates about its own decision, so your explorer reads like a real audit trail instead of a wall of hashes:

aml-screening-agent:cleared:no-match-ofac-sdn-list
kyc-verification-agent:verified:govt-id-liveness-check-passed
loan-underwriting-agent:approved:income-verified-debt-ratio-28pct

Your payload stays exactly what it always was — an opaque hash of your real, sensitive decision. The label is just metadata you choose to disclose, and it's just as tamper-evident as the hash itself, permanently bound to the same record.

A single agent with a clean decision function is a few lines of Python — genuinely that simple. It stops being that simple the moment you have several agents built by different teams, an orchestration framework where the "decision" isn't one obvious function call, or a specific compliance schema (EU AI Act Article 12, ISO 42001) your audit trail actually needs to satisfy. That's real integration work, not an afternoon.

Try the demo →

Under the hood

Real Java 21 / Spring Boot / Spring Cloud microservices — Config Server, Eureka service discovery, an OAuth2-secured Gateway, event-driven services over Kafka, containerized and deployed via Kubernetes/Helm. Every box below is a real, independently-deployable service, not a conceptual diagram.

Client POST /api/tx · Bearer JWT Keycloak OAuth2 / OIDC identity issues + validates JWTs API Gateway Spring Cloud Gateway JWT auth · Resilience4j ingest-service hash + publish Kafka event stream transparency-service single writer · sequencer real tamper-evident hash chain Prometheus · Loki · Tempo → Grafana dashboards metrics · logs · traces PostgreSQL chain records Eureka + Config Server discovery + centralized config for every service above Kubernetes / Helm hosts every service below GKE, templated Helm chart
Full architecture + source → Try the live demo →

Helps you comply with

Every AI-governance framework converges on the same requirement: prove what your AI did, when, and that the record wasn't altered afterward. That's exactly what Entropa's log gives you. See the full, honestly-labeled control list on the Trust & Controls page.

Bank Secrecy Act / FinCEN recordkeeping
Entropa provides the tamper-evident recordkeeping component that BSA/FinCEN recordkeeping rules require for decisions supporting your AML program — not a replacement for the program itself.
SR 11-7 model risk management
Federal Reserve/OCC guidance calls for documented, verifiable evidence of what a decisioning model actually did. Entropa provides that evidence trail — not the model validation or governance program itself.
EU AI Act — Article 12
Automatic, tamper-resistant logging and traceability for high-risk AI systems.
NIST AI Risk Management Framework
Govern and Manage functions call for documented, verifiable records of AI system behavior.
ISO/IEC 42001
AI management systems require maintained, controlled documented information on AI operation.
IAPP AIGP
Accountability principle: demonstrate, with evidence, what an AI system did.

Trust is proven, not claimed

Once an attestation is recorded on Entropa, it can't be silently altered or removed — not because we try hard, but because of how the log is built.

Isolated per customer
Every customer gets their own private, append-only Merkle tree. Two customers' histories can never physically occupy the same structure — structurally, not by convention.
Append-only, checkpointed
Each record is folded into a signed checkpoint over the log's current root. A later record can never rewrite an earlier one's proof — the math that makes the checkpoint valid depends on it.
Cryptographically signed
Checkpoints are signed and independently verifiable — every proof can be checked against the signed root, not just claimed.
Independently verifiable
Every record page runs and displays a real inclusion-proof and signature check — the actual cryptographic result, not just a claim about it.
Transparent to a fault
Every real incident — bugs, mistakes, architecture decisions we later reversed — is documented publicly with root causes and fixes, not summarized after the fact.

Pricing

This is a free portfolio/demo project — not a paid product. Nothing here costs anything.

Free — demo / test only

Signup runs a real Stripe test-mode checkout ($0, no real charge) purely to demonstrate the flow end to end. No real customers, no real billing, ever.

In a real deployment, a system like this would replace hiring the infrastructure, DevOps, and AI engineering to build and run a provably tamper-evident audit trail yourself — this project exists to demonstrate that exact engineering (Java/Spring microservices, Kafka, Kubernetes, real standards-mapped design: EU AI Act Article 12, NIST AI RMF, ISO 42001), not to sell it.

Frequently asked questions

What is Entropa?

Entropa is a post-quantum, tamper-evident Merkle transparency log for AI-driven decisions in regulated finance — built for banks and fintechs that need an independently verifiable record of what their fraud, KYC, and AML screening agents decided. Every submission is appended to the log and covered by a signed checkpoint. Anyone holding a receipt can independently verify their record is really in the log, and that the log hasn't been secretly rewritten, without trusting Entropa's own server code.

Do we already have to keep AML/KYC decision records — why do we need this too?

You already keep records. The question an examiner or auditor asks is whether those records could have been altered after the fact — and a database row your own team controls can't answer that on its own. Entropa doesn't replace your existing recordkeeping; it makes the specific records you route through it independently verifiable, so "trust our logs" becomes "here's cryptographic proof," without changing your AML program itself.

Is Entropa a cryptocurrency?

No. Entropa has no token, no mining, and nothing to speculate on. It's billed like a metered API — Stripe or Twilio style — not a coin.

Is Entropa decentralized?

No, deliberately not. Entropa is a single-purpose, tamper-evident transparency log, not a general-purpose decentralized blockchain platform. There's no smart-contract language and no wallet. An earlier version of Entropa ran real multi-validator quorum consensus in production; it was removed once it became clear it solved a multi-party-trust problem this single-operator project doesn't actually have.

Do you ever see our customers' data?

No — structurally, not just by policy. You compute a hash of the decision on your own systems before anything is sent to Entropa; we only ever receive that hash, never the underlying customer name, account number, or transaction detail. For vendor-risk review, that means there's no sensitive data on our side for a due-diligence process to have to protect against.

Is my data private?

Yes. You get your own private Merkle tree, isolated in storage and in the cryptographic structure itself. You can download your full attestation history whenever you want, directly from your dashboard.

How much does Entropa cost?

Nothing — this is a free portfolio/demo project, not a paid product. Signup runs a real Stripe test-mode checkout ($0, no real charge) just to demonstrate the flow end to end.

How is billing handled, and how do I cancel?

You're charged automatically on the 3rd of every month for that cycle's usage — no invoices, no bills sent for later payment. Your card stays securely on file with Stripe; we never see or store it ourselves. If you cancel from your dashboard's Your Data section, that same card is charged immediately for usage from the last 3rd up to the moment you cancel, and your account and entire private Merkle tree are permanently deleted right then. Download your attestation history first if you want a copy — deletion is immediate and irreversible.

Which regulations and frameworks does Entropa help satisfy?

Entropa's tamper-evident audit trail provides the specific recordkeeping/evidence component that BSA/FinCEN recordkeeping rules, SR 11-7 model risk management guidance, the EU AI Act (Article 12), the NIST AI Risk Management Framework, ISO/IEC 42001, and the IAPP AIGP body of knowledge each call for — not a replacement for the compliance program or standard itself.