How to volunteer for an AI agent pilot as a KYC analyst

Offer to be the pilot's evidence checker. Ask your KYC team lead to nominate you as a reviewer, bring examples of onboarding files where the documents looked complete but were not, and volunteer to decide every screening match and risk rating the agent drafts. Log each decision with its reason. That log is the case for your next role.

What part of KYC is a bank likely to pilot?

KYC work in retail banking is document-heavy and repetitive: checking identity documents, proof of address, business registration details and ownership structures, running names through sanctions and politically exposed person screening, drafting the risk rating, and chasing customers for missing items. Periodic reviews add a second stream of the same work on existing customers.

An agent can assemble the file, check it for completeness, run the screening, and draft a summary with a suggested risk rating. The onboarding operations manager owns the pilot, and financial crime compliance owns the policy. Under the FATF Recommendations, due diligence is risk-based, which means judgement about each customer stays central. The analysts who know where files go wrong are the ones the pilot needs. The guide to AI agent jobs and careers shows where this work is heading.

How should the decisions be divided?

1. Agent, depth 1: reads the application, the documents and the screening results. Data from an official company registry is signed. What the applicant states about their income or occupation is declared. An uploaded utility bill is external until verified.
2. Agent, depth 2: drafts the completeness check, the screening summary and a suggested risk rating with reasons.
3. You: decide each held file: approve, request more documents, clear or escalate a screening match, adjust the rating.
4. Agent, depth 3: once certified on the bank's own past files, sends standard requests for missing documents within agreed rules.
5. Agent, depth 1: flags every potential screening match, every complex ownership structure and every high-risk indicator.
6. Senior analyst or financial crime team: decides enhanced due diligence cases and true matches.
7. You: record a one-line reason for each decision.

How do you make the case to join?

1. Ask your team lead who is selecting reviewers and which customer segments the pilot covers.
2. Prepare three anonymised examples of files that looked complete but were not: a proof of address that did not match the application, an ownership chain that hid a controlling party, a screening match cleared too quickly by someone else.
3. Offer a specific role: deciding the held files and screening matches in one segment.
4. Show your quality record. Review seats go to accurate analysts.
5. Agree cover for your periodic reviews, then send the offer using the email template to volunteer for an AI agent pilot.

Which checks must never be skipped?

These are your red lines. Name them in your offer and own them during the pilot.

• No account opens with an unresolved screening match. Evidence: your clearance or escalation against each match in the append-only record.
• No external document counts as verified on its own. An uploaded bill or statement is checked against a stronger source before it supports a decision. Evidence: your verification notes.
• High-risk indicators trigger enhanced checks. Evidence: the referrals to the senior team.
• Ownership is understood before approval. If the controlling party is unclear, the file waits. Evidence: your hold reasons.

What should your log show at the end?

The files you decided and the reasons, the screening matches you cleared and escalated, ratings you changed and why, documents you rejected, and the recurring problems that stopped after a source or rule was changed. Use the workspace's filed verdicts. If periodic reviews are in scope, keep them separate in your log, because the errors differ: onboarding files fail on missing evidence, while reviews fail on changes nobody noticed, such as a new director or a new country of operation.

That is the outline of a role such as onboarding operations analyst, KYC quality reviewer or financial crime operations specialist. Titles differ across banks; read the duties for review, exceptions, approvals and escalation. How a KYC analyst can move into AI operations and what jobs a KYC analyst can get with AI agent skills map the options.

What mistakes do KYC volunteers make?

• Trusting completeness as correctness. A full file can still be wrong. Completeness is the agent's job; truth is yours.
• Clearing matches on name alone. Check the other identifiers the policy requires.
• Not feeding back. If the agent keeps misreading one country's identity format, report it so the source changes.
• Working in isolation. Fraud and compliance colleagues often sit on the same pilot; see the playbooks for fraud analysts and compliance analysts.

Write down the reasoning behind your hardest decisions as you go, too. Interviewers for oversight roles tend to ask how you handled an ambiguous file, and a real example, described without customer details, is far stronger than a general answer.