US Tax & Legal Research
Grounded answers with exact citations · Graph RAG
Try

Ask a legal or tax question in plain English. Every answer is grounded in the source documents, with clickable citations to the exact document, section, and page.

Pick any of the 100 corpus documents to get a concise, faithful summary of its key provisions.

Graph RAG — find which documents cite or reference a given Act, statute, or case. Something pure semantic search can't answer.

Try

Enter an Act, statute, or case name to traverse the citation graph and see every corpus document that references it — grouped by type.

Every question is graded from a full evaluation run — run any one to confirm it live against the hand-authored ground truth.

95% / 99%
retrieval top-1 / top-5
0.96
faithfulness
91% / 98%
correct / correct-or-partial
100% / 0%
grounded / hallucinated
Download CSV
100
documents
6,289
chunks indexed
95% / 99%
retrieval top-1 / top-5
0.96
faithfulness
100%
grounded / cited
0%
hallucination
System architecture: indexing pipeline, shared vector/keyword/graph stores, and the live query path to a grounded, cited answer

Hybrid retrieval (vector + BM25, fused with Reciprocal Rank Fusion) plus Graph RAG over a Neo4j citation graph, with grounded generation via Gemini 2.5 Pro — every answer carries mandatory citations (document · section · page) and passes a citation-validation guard.

Evaluation Report

End-to-end on a hand-authored 100-query golden set (25 per category: Acts / Court Judgments / POV / Tax). Answer model: Gemini 2.5 Pro.

Retrieval accuracy

CategoryTop-1Top-5
Acts100%100%
POV (commentary)100%100%
Court Judgments96%96%
Tax Documents84%100%
Overall95%99%

Faithfulness score

Overall performance on the golden set

Retrieval (Top-1 / Top-5)FaithfulnessCorrectnessGrounded / Halluc.
95% / 99%0.9691% / 98%100% / 0%

Observations, limitations & improvements