Worked examples

Each example stands alone. Save the JSON and Rego files shown on the page, choose Python, TypeScript or cURL, and run the client with your API key. All requests use https://api.blatent.ai.

Example	What it demonstrates	Extra context needed
Classify a message	Descriptions, enums and a boolean decision	None
Extract a booking	Text, identifier patterns and date-time formats	None
Connect output fields	Conditional fields and legal combinations	None
Apply permissions	Schema conditions plus Rego checks against scopes and state	Application-supplied permission snapshot
Check totals	Extracted values with exact arithmetic consistency	None
Extract facts and calculate	Variables extracted from prose and checked arithmetic steps	None
Order dates	Relationships between timestamps and an allowed window	Window supplied as state
Constrain a poem	Exact text counts and policy repair feedback	None

Examples distinguish the model's interpretation from what the schema and policy enforce. Outputs marked illustrative explain the response shape. Observed results are identified separately; the poem also demonstrates policy-guided search. See evaluation for testing your own inputs.