Watch ClauseLens Run
A hybrid-retrieval RAG pipeline for contract intelligence: structural chunking, BM25 + dense fusion, citation-validated answers. Pick a question — this replays a real recorded run of the open-source pipeline.
clauselens · pipeline replay · corpus: 3 contracts / 24 chunks
90.7%
hit-rate @8 · 43-question golden set
1.5ms
mean retrieval latency
100%
answers carry citations
0
API keys needed for this demo
What you're watching, honestly: these are recorded runs of the real
clauselens-rag pipeline against its sample corpus —
real queries from the benchmark golden set, real retrieved chunks, real latencies from
benchmark_results.json.
Nothing here calls a model at page-load; that's the point. Clone the repo and run the same pipeline locally with or without an API key.
The pipeline
contracts (.txt/.pdf) ──▶ structural chunker ──▶ ┌────────────────────┐
│ BM25 index │
│ dense vectors │──▶ Reciprocal-Rank Fusion ──▶ top-k chunks
└────────────────────┘ │
query ──▶ query rewriting ──▶ hybrid retrieval ────────────────────────────────┘
▼
citation validator ◀── synthesis (Claude / extractive fallback)
│
▼
answer + clause-level citations (rejected if citations don't verify)
Five verticals, one architecture
Each fully specified: architecture, costs, pricing, go-to-market. ClauseLens is built; the rest are queued.