BAML
A glue language positioned between AI agents and business code. It uses TypeScript-like syntax to define LLM output format specifications and automatically repairs JSON formatting errors produced by models at runtime.
Timeline (last 90 days)
Agentic GraphRAG's entity extraction is driven by a user-defined schema, with local extraction using GLiNER and structured LLM extraction handled by BAML on selected chunks.
BAML is a domain-specific language for defining and calling LLM functions; once compiled it generates type-safe client code that enforces schema-conforming JSON output in structured extraction scenarios.
Entity extraction in Agentic GraphRAG can use a locally running GLiNER model, or BAML can be used for structured LLM extraction on selected chunks.
GPT-OSS-120b has nearly a 50% chance of making JSON formatting errors when used outside of BAML.
All Facts (4)
Agentic GraphRAG's entity extraction is driven by a user-defined schema, with local extraction using GLiNER and structured LLM extraction handled by BAML on selected chunks.
50%UnverifiedBAML is a domain-specific language for defining and calling LLM functions; once compiled it generates type-safe client code that enforces schema-conforming JSON output in structured extraction scenarios.
50%UnverifiedEntity extraction in Agentic GraphRAG can use a locally running GLiNER model, or BAML can be used for structured LLM extraction on selected chunks.
50%UnverifiedGPT-OSS-120b has nearly a 50% chance of making JSON formatting errors when used outside of BAML.
50%