guides

Cookbooks

Each recipe is a worked example with a runnable request in curl, Python, and TypeScript, the actual output we received when we ran it, and a note on when to trust the result. Recipes cover three models: Decider for typed decisions with probabilities (/v1/systemone; the Decider recipes were recorded with Decider 1.0, and decider-1 now serves 1.1, so expect small differences in the numbers), Restyler for rewriting AI-sounding prose (/v1/restyle), and Fanout for predicting the web searches an AI assistant is likely to run (/v1/fanout). The Restyler and Fanout recipes were recorded with their 1.0 versions; restyler-1 and fanout-1 now serve 1.1, so outputs will differ a little. Outputs are shown unedited, including cases where a model got something wrong.

Decider

• Route a support email in one call · Support inbox triage
• Act only when the model is sure · Confidence-gated actions
• Flag a phishing email · LLM guardrails
• Drop retrieved passages that do not help · RAG passage filtering
• Hold a reply that promises money · LLM guardrails
• Pick an agent's first tool · Agent routing and skill selection
• Ask many questions in one call · Structured extraction
• Decide what an agent should remember, before Mem0 stores it · Agent memory with Mem0

Restyler

• Restyle a draft, then check the numbers · Content editing

Fanout

• Turn a buyer's question into a content brief · Generative engine optimisation
• See which rivals an assistant pulls in · Competitive research