Developers
Build with the
Semarize API
Turn conversations into structured, machine-readable signals. Send conversation content and a Kit ID. Get deterministic JSON with scores, flags, evidence, and metadata.
How it works
Content in. Signals out.
The Semarize API accepts conversation content plus a Kit ID and returns structured evaluation results. Every signal is typed, evidence-backed, and versioned.
Core concepts
Bricks, Kits, and Runs
Three primitives power every Semarize evaluation.
Bricks
A Brick is a single semantic check. It evaluates one specific thing about a conversation - a boolean flag, a numeric score, a categorical label, or an extracted text value. Each Brick has configurable output fields: value, confidence, reason, and evidence.
Kits
A Kit bundles related Bricks into a reusable evaluation framework. Run a Kit against a conversation and every Brick inside it executes. A Kit with 8 Bricks returns 8 structured signals.
Runs
A Run is one execution of a Kit against one conversation. Submit a conversation and a Kit ID to the API. Get results via polling, sync mode, or webhooks. Semarize processes all Bricks and returns structured JSON with values, reasons, and evidence spans.
Base URL
All requests use a single base URL
All API endpoints are versioned under /v1. Authentication is via Bearer token in the Authorization header.
Documentation
Jump to a section
Quickstart
Get your API key and make your first run in under 5 minutes.
API Reference
Complete endpoint documentation - Runs, Webhooks, and error handling.
MCP Reference
Connect AI agents to Semarize via the Model Context Protocol.
Automations
Connect Semarize to Clay, n8n, Zapier, and Make with step-by-step flows.
Security
Authentication, key management, data handling, and access control.
Ready to build?
Get your API key and start extracting structured signals from conversations in minutes.