How to move from KYC analyst to AI agent reviewer

KYC analysts are natural reviewers of AI agents because their job is deciding how much to trust a piece of evidence. Agents collect documents, screen names and draft risk ratings; reviewers decide the matches, mismatches and enhanced due diligence cases. Learn how governed agents weigh sources, practise reviewing their files, earn a certificate employers can check and apply.

Is KYC experience enough to become a reviewer?

Mostly, yes. Customer due diligence is already a review discipline: you check what a customer declares against documents and independent sources, decide whether it is enough, and escalate when it is not. The FATF Recommendations, the international standard behind most anti-money laundering rules, set out this risk-based approach, and you apply it every day.

What you need to add is the agent's side. An agent in retail banking onboarding can request documents, check expiry dates, run names through screening, compare addresses, draft a customer risk rating and prepare a file for approval. You need to know how far it may act, which sources it may rely on and how its verdicts are recorded. The guide to AI agent jobs and careers sets out the wider set of roles.

Why does provenance matter so much to KYC reviewers?

Governed agents tag every fact with where it came from. Sealed facts come from a system of record, such as the bank's own ledger. Signed facts come from a named party who answers for them, such as a verified issuer of a document. Declared facts come from the customer. External facts arrive from outside, such as a web page or an emailed letter.

That is your existing thinking about evidence, made explicit. A customer's stated source of funds is declared. A bank statement from another bank, received by email, is external unless it can be verified. A certificate issued and signed by a verified issuer is signed. As a reviewer, you check that the agent weighed each fact correctly. An agent that treats a declared source of wealth as proof has made exactly the mistake you would catch in a junior analyst's file.

What are the steps?

1. You, weeks 1 to 2: learn depth, verdicts and provenance. Map each provenance class to the documents you handle.
2. You, weeks 2 to 4: re-read ten files you escalated. For each, write what an agent would have concluded and where it should have held. Note the source it might have over-trusted.
3. You, weeks 3 to 5: offer to review onboarding agent output. Banks often pilot agents on document collection and screening first.
4. You, weeks 4 to 8: complete a practicum. Two weeks of review decisions to a measured standard, on a live or simulated queue.
5. You, weeks 6 to 9: earn a certificate employers can check.
6. You, week 8: rewrite your CV. "Completed 25 CDD reviews a day" becomes "reviewed onboarding files for evidence quality; held cases relying on unverified source-of-funds claims; escalated PEP and sanctions matches with written reasons."
7. You, weeks 8 to 14: apply. Target onboarding review, financial crime operations and AI operations roles. Adapt the cover letter template for KYC analysts.

What will you review?

An illustrative retail bank onboarding queue:

What evidence matters to employers?

Bring your practicum decision log, your assessor's sign-off and a certificate number employers can look up. Add one short note on provenance: three facts from a real (anonymised) file, the class you would assign each, and why. That shows instantly that you understand the core idea of agent review.

If you have worked periodic reviews, mention them separately. Periodic review is where agents save the most time and where stale evidence slips through, so employers value reviewers who know what changes in a customer's file over a few years.

How do you know it is working?

Your practicum accuracy holds over both weeks without a missed red line; opening an account on a confirmed sanctions match would be one. Interviews shift from screening tools to judgment about sources. Financial crime operations and onboarding teams start replying with review roles, not only analyst roles.

Where does it usually go wrong?

• Over-escalating. KYC teams are trained to escalate when in doubt. Reviewers who hold everything fill the queue; hold what needs a person.
• Assuming the agent screens like a person. It may handle spelling variants differently. Check how it matched, not just whether it did.
• Forgetting the customer. A request for documents can be accurate and still confusing. Review tone too.
• Leaning on tool names. Screening systems differ by bank; your judgment is what transfers.

Related moves are described in moving from compliance analyst to AI agent reviewer, moving from fraud analyst to AI agent reviewer and how a compliance analyst can move into AI operations. The broader KYC route is in how a KYC analyst can move into AI operations.