Virage
Chunking and retrieval engine for AI/RAG pipelines
Virage turns codebases and documents into high-quality context for LLMs: a chunking and retrieval engine with an open-core CE (community, tree-sitter/markdown/etc. chunkers) and an Enterprise Edition adding XLSX/PPTX/HTML/ePub/reStructuredText chunkers, licensing, and enterprise-scale features. Foundation-stage — see maturity note below before treating anything here as a completed sale or production deployment at a customer.
Maturity
Foundation stage (of a Foundation → Adoption → Monetisation arc). Commercial infrastructure (license server, SDK, portals, checkout) exists, but there are zero paying customers and no sales outreach has been executed yet.
Roadmap highlights
- CE community chunkers in progress: Jupyter notebook, OpenAPI/Swagger, Confluence HTML export
- EE observability shipped: tracing/logging with syslog and OTLP-over-HTTP exporters
- Core retrieval-quality work planned: query-aware context assembly, HyDE, NER enrichment, keyword extraction
- EE chunker hardening shipped: XLSX, PPTX, HTML, with fixture-based integration tests