Show HN: Tessera – an AI agent that refuses to answer without evidence (github.com)

Tessera has emerged as a groundbreaking AI agent that prioritizes evidence-based responses and actions, operating within an open, deterministic framework. Unlike typical AI systems that might blend disparate sources and present unverifiable claims, Tessera ensures every answer is supported by a traceable proof, refusing to generate responses where evidence is lacking. This approach significantly enhances trust in enterprise AI applications, tackling the pervasive issue of misinformation stemming from weak source validation and composite data. The system integrates diverse data sources—ranging from structured databases to unstructured documents—into a unified knowledge graph, allowing for precise questioning through a conversational interface. Each claim made by Tessera is backed by citation paths, while its innovative benchmarking harness quantifies trust through measurable metrics like faithfulness scores. Furthermore, it features a rigorous approval process for any actions, ensuring that changes are only executed with clear, verifiable evidence. This holistic emphasis on accountability and transparency positions Tessera as a potentially transformative tool for organizations looking to implement reliable AI solutions in their operations.