How we use AI

Product documentation for every customer-facing AI surface — what is measured (deterministic scores, metering, Admin API cost), what is generative (compliance chat, questionnaire and policy drafting, the AI Security Audit, and summaries of vendor attestations or contract text you paste in), and what is extractive/classifier (trust buyer chat is not a generative bot). Documented from code. Not legal advice.

Last verified September 27, 2026

How does ShipReady Metrics use AI?

Eight customer-facing AI surfaces: generative compliance chat, extractive trust-buyer classifier, deterministic score graders, telemetry ingest, metering, cost, provenance, and UI labels. From code.

How does compliance chat use AI?

Compliance chat: READ-ONLY generative Anthropic Q&A grounded on the org readiness view via createOrgAiClient. No tools, no verdict path. Flag- and org-gated.

How does the public trust buyer chat work?

Public trust buyer chat: deterministic FAQ first, then citation-gated Anthropic classifier (topic token only). Grounded on published PublicTrustCenterView. Not a generative bot.

What AI features are measured, not generative?

Deterministic AI ROI, AI Readiness, and Agent Health graders; AI-generated badges; agent telemetry ingest (no prompts); Anthropic Admin cost; usage metering; model provenance registry.