Tessera SDR · Phase 1 build

Stop AI sales agents from leaking pipeline.

The accountability layer for every AI-generated sales touch. Tessera watches what your agents are about to send, catches the moments they would have hallucinated, contradicted, or violated an opt-out, and intervenes before bad messages reach the buyer. Read-only audit by default — observe first, intervene when ready. Public landing at tesseraai.io.

Tessera knows when the Zap should NOT run.

Observe

Read-only audit of every AI-generated sales touch — no vendor calls until you flip the switch

Decide

Allow / block / pause / escalate / route — eval-grounded judgment, not a workflow constructor

Prove

Incident Ledger logs every prevented leak — pipeline value at risk surfaced at the top

Health

  • · lib/types/lead-state.ts — shipped
  • · lib/vendor-adapters/{interface, reply-io, conversica, registry}.ts — shipped
  • · lib/decision-engine/{rules, classifier, engine, orchestrator, meeting-flow}.ts — shipped
  • · lib/eval/{persona-generator, roleplayer, judge, harness, llm-clients, simulation, loss-analyzer}.ts — shipped
  • · lib/data/{dashboard, prevention, meetings, postmortems}.ts — shipped
  • · app/api/webhooks/{reply-io, conversica}/route.ts — shipped
  • · supabase/migrations/0001-0007 — shipped
  • · methodology/objections/{saas-pricing, healthcare-pricing-compliance}.yaml — shipped
  • · tests: 159 passing across 16 files