iServ

iServ Governance Hub

An AI platform that interviews a company about its security posture, then drafts its governance documents with a source map behind every claim.

Status: LiveShipped: June 2026
commits in 6 days, overnight autonomous loop
governed document templates
lettered acceptance checks run per commit
layer safety gate on every inbound message

01 The story

Small companies need a written security program, an incident response plan, and policy bundles, and they usually get a template with their name found-and-replaced into it. This hub interviews them instead: a conversational security advisor tracks a structured profile to 85 percent completeness, a second model extracts fields from every message, and a safety gate classifies inbound text before anything persists. Generated documents carry a source map: every claim traces to a profile field with a confidence flag, and a credentialed human reviews before anything publishes. Versions are immutable with full provenance.

25,783 lines of code

02 What went into it

  • Conversational discovery interview with live profile extraction
  • Three-layer safety gate: pattern check, classifier, audit
  • Document engine with per-claim source maps and immutable versions
  • Six templates: WISP, incident response, breach notification, SOC 2 and ISO readiness, roadmap, policy bundle
  • Regulatory branches: MA 201 CMR 17.00, HIPAA clocks, SOC 2 and ISO paths
  • Credentialed review gate before any document publishes
  • Customer readiness snapshot scored 0 to 100 with domain subscores
  • Acceptance harness that boots its own server and checks end to end

03 Claude in production

Security-advisor interviewPer-message field extractionSafety classificationGoverned document generationWebsite pre-fill drafting

Every AI call runs server-side under the Lumen governance layer: no client keys, cost telemetry on every call, and a human confirming every consequential write.