Ever wondered how a fast‑response police officer becomes an AI‑first CEO shaping the future of audit and forensics? We sit down with Lem Chin Kok to trace a rare career arc—from crime scenes to commercial investigations to building tech‑led forensic teams—then dig into the practical ways AI transforms how accountants work. Lem explains, in plain terms, the difference between classical predictive models that spot high‑risk transactions and generative AI that removes drudge work, and why both are essential when your ledger runs to millions of lines.
We explore the moment an unsupervised model surfaced ten “most unusual” entries from tens of millions—and nine proved to be real frauds later charged in court. The secret wasn’t magic; it was feature design grounded in accounting records and fraud behaviour. That insight anchors a bigger theme: domain expertise plus data fluency beats black‑box buzzwords. Lem makes the case for a “third language” for professionals—programming—so auditors and finance teams can shape models to client context, express risk hypotheses as features, and build repeatable, explainable tests that scale beyond random sampling.
The conversation widens to enterprise AI. Consumer tools can be brilliant yet random, which breaks policy‑driven workflows. Lem introduces the “Gen AI twin,” a policy‑aware system that executes end‑to‑end tasks for finance, compliance, or operations, producing consistent outputs aligned to organisational procedures while keeping humans in the loop for judgement and sign‑off. We also touch on building an applied forensic qualification spanning digital forensics, AML, sanctions, and investigations, and why partnerships with public agencies help keep training close to reality.
If you care about fraud detection, audit quality, data analytics, or how to deploy AI responsibly in the enterprise, this conversation offers clear frameworks and next steps. Subscribe, share with a colleague who needs to hear it, and leave a review with the one capability you think every accountant should learn next.