Grounding LLMs with LexBharat AI

Why and how to ground model answers in retrieved, authoritative statutory text.

A model answers better, and more safely, when it is given the text it must answer from. Retrieve sections first, pass their text as context, and ask the model to cite the actId and citation it used.

{ "query": "…", "corpusType": "USC", "limit": 5, "includeBody": true }

• Quote only what was returned.
• Report currency (goodLawStatus) alongside the answer.
• Say so when nothing is found; do not fill the gap.