Eshaan Mathakari online · NYC

Rulebook Radar

Eight AI agents turn a regulatory document in any language into a reviewed change record.

Status
Proof of concept
When
Sep–Oct 2026
Role
Built at work
Area
Professional work

In plain English

Rulebook Radar is a proof of concept for managing regulatory change. Eight AI agents translate a regulatory PDF, check that it is in scope, fill in a structured change record with the rule citations it affects, review the full text of each citation and check their own work. Low-confidence cases go to a person, accuracy is measured against benchmark records, and signed-off reports export as PDF, Word or CSV.

How it works

  1. Translate the regulatory PDF, whatever language it is in.
  2. Check that the document is in scope.
  3. Fill in a structured change record with the rule citations it affects, then review the full text of each citation.
  4. Run QA on the agents' own output and send low-confidence cases to a person.
  5. Measure accuracy against benchmark records, and export reports as PDF, Word or CSV after expert sign-off.