Repass Intelligence
What your company knows is scattered. The work is connecting it responsibly.
Repass Intelligence builds the layer between a company's existing systems and the people who need answers from them. Documents, spreadsheets, financial records, calls, and the rules that govern them, connected by an AI agent that cites its sources and keeps sensitive data where it belongs.
I work with a small number of companies in workers' compensation, PEO, and professional services, integrating what they already run rather than replacing it. The AI is the integration layer, not the product.
How it fits together
Each kind of information lives in the store built for it. One agent works across all of them and says where each answer came from.
The right store for each kind of data
Prose goes into retrieval systems that find and cite it. Numbers stay in databases that can sum, join, and audit them. A language model is never asked to remember a figure it could look up.
Evidence before prescription
Before anything is automated, the current process is mapped from what the systems actually recorded and checked with the people who do the work. Recommendations come after observation, one piece at a time, each one measured.
Discretion by design
Client work is not described in public, and sensitive documents are processed on GPU hardware I own rather than sent to a third-party service. Cloud models are used only where the data allows.