Tessera

The agent can only say what it can prove — and only do what you approve. An open, deterministic evidence layer for AI agents: claim-level provenance, principled refusals, approval-gated actions with receipts — and faithfulness as a number in CI, not a vibe.

Try it live: https://robert-vetter-tessera.hf.space

Tessera ingests heterogeneous enterprise data (tables, spreadsheets, documents, logs, tickets), resolves the entities scattered across those sources into a single knowledge graph, and answers questions through a conversational interface in which every individual claim is grounded in a traceable path back to the source records that support it. A built-in benchmark harness turns "do you trust this AI?" from a feeling into a measured faithfulness score.

Start here

• Benchmark — The Faithfulness Floor — evidence-gated vs. the same engine ungated; CI-pinned; how to attack it.
• Demo & Hosting — the live surface, the 3-minute script.
• Pilot in a Day — grounded answers on your own repo or CSVs in under 30 minutes.
• Project Brief — motivation, the gap, vision, principles, scope, and what "excellent" means.
• Capabilities — the four pillars at feature level and what each must guarantee.
• Technical Write-up — the whole story in one sitting: approach, the recorded results trail, limitations, lessons.
• Roadmap — the phased build plan across ~3–4 months.
• SAP Alignment — how the work maps onto SAP's teams and platform.
• Deployment (SAP) — the AI Core / GenAI Hub / HANA Cloud path, and why local mode is the default.

Build and operate

• Engineering Handbook — workflow, tooling, quality gates, and anti-drift discipline.
• Setup — from docs to a running, gated project.
• Status — the living session journal.
• Architecture Decision Records — why it was built this way.

---

This site is generated from the docs/ folder. The source lives on GitHub.