What it does
Its six replay claims illustrate the workflow, not validated financial risk. Live mode sends claim and receipt text to TypeSafe; the example records decisions but makes no payments.
Apps & data pipelines
nl-jev-decision-gates-expense-policy
An expense-policy example combines Jev’s receipt judgments with fixed checks to approve, reject, or send claims for human review.
Bookmark: nl-jev-decision-gates-expense-policy
Keep this in your collection.
Leave a noteWhat would you try with this?
: nl-jev-decision-gates-expense-policy
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
How you can use it
A developer can adapt this project to review employee expense receipts. The software uses regular code to check basic rules and reject duplicate requests outright. Then, your developer uses an API key to connect the software to TypeSafe. This key is a unique code that links your application to the outside AI service. The AI reads the receipt text to spot unclear business reasons or prohibited items.
The system waits for a person if a claim exceeds limits or the AI is uncertain. The software does not actually pay any money. Your developer can run a built-in test with six sample expenses. This lets you see the process without connecting the live service.