bossy v0.1.0 — Initial Release
A self-hosted, privacy-first ChatBI platform. Ask questions in plain language and get SQL, tables, and charts over your own databases — without your data ever leaving your infrastructure.
Highlights
- Natural-language → SQL over PostgreSQL, with read-only execution enforced.
- Hybrid retrieval (vector + BM25 + cross-encoder rerank) over metadata, semantic definitions, and historical SQL.
- Semantic layer that deterministically maps business terms (metrics/dimensions) to columns.
- SQL safety guard — syntax/field/permission validation plus row-count and scan-cost limits before any query runs.
- Automatic visualization — derives an appropriate chart spec from the result shape.
- Agent pipeline with 8 specialized LLM roles (intent, planning, SQL gen/rewrite, insight, …), each independently tunable.
- One-click analysis — generates a business overview and improvement suggestions across the whole metadata catalog.
- Multi-datasource routing — a single question can be routed across multiple business databases, picking the right source and join path.
- Bring-your-own-LLM — run fully local via Ollama (zero data egress), or connect any OpenAI-/Anthropic-compatible endpoint. Keys are encrypted at rest.
- Built-in metadata workbench — curate tables, relationships, and metric definitions through a UI, no YAML required.
- Realistic demo dataset — a synthetic banking/financial dataset (7 domains) generated locally with Faker, ready to explore via
make init.
Docs
Full Changelog: https://github.com/ProjAnvil/Bossy/commits/v0.1.0