Participated in public demo 2601 with Reglens

The problem [00:55] Compliance teams at banks, money transfers, credit unions, and law firms spend 60-70% of their time reading, searching, and re-explaining regulations rather than making judgment calls. A single regulatory update can take days to weeks to interpret across the U.S., Europe, Asia, and other jurisdictions, and senior staff end up used as human search engines. 85% of the work is retrieval and recall, only 15% is true judgment. What they built [02:05] RegLens (RegWatch AI / RegGuard AI) is an AI compliance triage agent with a confidence-sorted review queue: each regulatory update gets an AI recommendation, confidence level, reasoning, and source citation with inline quotes from the actual regulatory text, so analysts can approve, correct, or escalate it. A Trends & Analytics page tracks agreement rate rising and correction rate falling over time, and an Attestations page gives regulators an audit trail of signed-off batches. Trust features include a blind review mode (analysts decide before seeing the AI's recommendation), consistency checks that flag contradictory overrides, and a frozen audit trail (analyst name, timestamp, prompt version, model version). What happened [04:20] Scored 10/10 on 15 stress-test cases across 4 categories (false positive/negative traps, ambiguous cases) informed by a Morgan Stanley compliance manager; a second AI model (ChatGPT) was used to gap-check the design and surfaced 20 additional risks not previously considered, including bias and institutional-memory contamination. In production, AI agreement rate is 88% with an 8% correction rate. The ask [06:38] Looking for collaborators who work in or are curious about compliance, fintech, or AI-regulated industries.