Because we set no identifiers, we genuinely cannot tell a new visitor from a returning one. Most analytics vendors treat that as a limitation. We treat it as the spec.
These aggregate breakdowns come from our GoatCounter dashboard and are updated periodically by hand — the all-time total above is live.
This is the feedback loop the triage runs on: practitioners submit stories of AI meeting financial systems, we verify and distill them, and the ones that generalize become scanner rules. A fourth distillation is written and awaiting a ship decision. The engine currently runs 37 rules, and 5 of them trace directly to field experience rather than vendor documentation. Full disclosure, in the spirit of this page: every story so far is the founder's own, told from scar tissue. Submit yours and change that.
Two ways to verify every number above. The data file your browser just loaded is public and refreshed hourly by an automated job — anyone can fetch it:
https://aiia-risk.com/data/daily.json
And our entire GoatCounter analytics dashboard is publicly viewable — the raw day-by-day, referrers and all, no login needed:
https://aiia-risk.goatcounter.com
That's the posture we bring to client work: don't ask people to trust the report — hand them the raw data. If you want this level of verifiability applied to the AI agents touching your financial systems, that's exactly what we do.