Thomas Geater
I write about what artificial intelligence is doing to financial-crime compliance: where generative AI breaks KYC controls, and where agentic systems have started running pieces of the compliance function themselves. Mostly I'm interested in what that asks of the people still accountable for it. The Typology Files is where I publish that work.
I graduated from the University of Brighton with a BSc in Business Management with Finance (2019–2022). The question that degree kept putting in front of me was whether what's on paper matches what actually happened, and AI has made that a good deal harder to answer. Onboarding documents a model produced in seconds. Due-diligence research an agent drafted rather than a person. Alerts a system triaged before anyone looked at them.
I read the primary sources rather than summaries of them: the MLRs, POCA, FCA guidance, the EU AI Act. I also run agentic AI tools against real compliance tasks before I write about them, and I publish what holds up.
Focus areas
Generative AI Fraud · Agentic AI in Compliance · AI Model Risk · Customer Due Diligence · Enhanced Due Diligence · Sanctions Screening · Beneficial Ownership Analysis · SAR Drafting
Selected work
- AML transaction-monitoring dashboard. JavaScript and Chart.js over the PaySim synthetic dataset of 6.3 million mobile-money transactions, laid out the way a screening queue is: confirmed fraud by transaction type, volume across 743 hourly steps, highest-risk cases surfaced first.
- AI Prompt Builder. A structured prompt generator for KYC, EDD, SAR and alert-triage tasks, built around the rule that an AI draft is a first pass a human has to verify. Runs entirely in your browser.
- The Typology Files archive. Long-form pieces on financial-crime typologies and AI in compliance, each worked up from the primary sources.
- GitHub. The code behind the above.
I work independently on this and take on advisory and writing engagements alongside the site. If it's a live problem where you are, I'm glad to talk about either half of it. Based in Brighton, working remotely.
Thomasdgwork@gmail.com