GovToKnow
Citation-grounded RAG assistant for municipal government documents
One example subsystem from this build. In most engagements we deliver the entire stack top to bottom — application code, Docker, infrastructure, deployment, and scaling — all in-house. Where a client already has platform teams in place, we work alongside them.
Project Summary
A citation-backed RAG assistant that helps residents get answers from their municipality's ordinances, codes, and official documents, with every answer grounded in inline citations to the source text. Built with Polity Inc. on Claude via AWS Bedrock, GovToKnow combines multiple chunking strategies, several search modes, and an adversarial test suite with groundedness and citation checks. Currently in demo stage: try the HappyTown demo municipality at govtoknow.com/happytown.
Our role: AI/Backend Engineering
What We Did
- Citation-grounded municipal RAG: every answer cites the ordinance or code section it came from
- Four chunking strategies for heterogeneous government documents (ordinances, codes, minutes, resolutions)
- Adversarial test suite with groundedness and citation checks gating every pipeline change
- Public HappyTown demo municipality at govtoknow.com/happytown
Want this in your own workflow?
Tell us which process you'd start with. We map it, install the loop, and prove it before it widens.
Plan Your Rollout