From policy PDFs to policy decisions
A rules first agent that tells up to 10,000 people who approves what, with the exact policy clause attached.
A large listed manufacturer with revenue above US$1 billionWhat the client was up against
The procurement rulebook spanned 70+ SOPs, a delegation of authority matrix per vertical and a capex policy, about 340 files in all. Requesters and approvers searched by hand and many questions ended with a few process experts. An earlier document chatbot pilot was dropped over running cost.
How we approached it
- Changed the design question from reading documents on every query to governing decisions
- Extracted policies once into reviewed, versioned decision tables
- Let the AI model interpret the question and explain the answer
- Kept documents as proof, with a citation to the exact clause
What we built
A Cortex policy agent that answers who approves what, which thresholds and quote rules apply and when a committee must sign off. It runs in the client's own Azure environment with its own model keys and a full audit trail.
The multi agent team behind it
The outcome it is designed for
This is live work without published results yet. We show the problem and the build, not numbers we have not measured.
Looking for a similar solution?
Tell us your version of this problem. We will walk you through what we built and what a first pilot would look like for you.
