About us

Risk infrastructure, built by the people who had to use it

FIN-TECHAI builds the risk layer that on-chain finance was missing: four production APIs for screening, credit, settlement routing and treasury reconciliation. We started because we were the compliance and treasury team stitching three vendors together with a spreadsheet, and we could not get a straight answer out of any of them.

Thesis

Explainability is the product

A risk score you cannot decompose is not a control — it is a liability you have outsourced. When a regulator, a bank partner or your own board asks why a customer was frozen or a loan was priced the way it was, "the vendor's model said 71" is not an answer.

So every number FIN-TECHAI returns comes apart. Weighted reason codes, hop distances, the exact transfer paths, the block height it was computed at. You can disagree with our weighting and change it. You can show your auditor the path. That is the whole design, and it is why our responses are larger and slower than a competitor returning a colour.

The second commitment is stating coverage. Every response carries what we did and did not see. A vendor that implies universal coverage is selling a blind spot with a confident face on it.

What we will not do
  • Take custody of funds or request withdrawal permissions
  • Return a score we cannot decompose for you
  • Claim coverage of a chain we do not ingest
  • Sell a compliance guarantee — that is not a thing anyone can sell
  • Put explainability behind a higher pricing tier
Where we are
Entity
Fin-techAI Foundation
Team
19 people across engineering, data and compliance
Working
Remote-first, hubs in London and Singapore
Founded
2024
Products live
Four modules in production
How we work

Four habits that shape the product

Backfill before you buy

Every evaluation starts by running your own resolved history through the API. If the result does not convince you, we would rather lose the deal than sell you a year of something that did not work in the test.

Ship the limitation with the feature

Each module page on this site has a section on what it cannot do. Those sections are written by the engineers who built the module, not by marketing, and they are the sections customers quote back to us most.

Reason codes are versioned

A scoring change that moves your alert volume is a breaking change. New reason codes are additive, weights are versioned, and you pin a version until you choose to move.

Coverage stated in-band

Every response carries its own coverage and the block height it was computed at. You should never have to read a marketing page to know whether an answer was complete.

Team

Who builds this

Nineteen people: chain data engineering, applied ML for graph clustering, and a compliance function with actual regulated-entity experience rather than a policy template. The compliance lead has right of veto on any claim this website makes about what a score means.

We are deliberately not publishing headshots and titles yet. When we do, it will be with real names and real accountability attached — the placeholder team grid is one of the least honest things a young infrastructure company can put on its site.

Hiring: chain data engineering and applied graph ML. Tell us what you have built.

Company timeline
2024Founded. Sentinel screening built first, for our own internal use.
2025 H1Sentinel to production on Ethereum and Tron. First exchange customers.
2025 H2Solana ingest. CreditGraph and PayRoute to production.
2026 H1Ledger Mind to production. SOC 2 Type II audit opened.
2026 H2Whitepaper published, $FTAI utility design finalised, VPC deployment available.

Come and stress-test the claims

The fastest way to judge an infrastructure vendor is to run your own data through it. That is what a demo with us is.