Skip to main content
Pi is an Agentic Infrastructure Intelligence Platform that gives you a single, consistent API surface across every major intelligence vertical. Whether you’re building brand-aware applications, AI-generated ad campaigns, real-time voice agents, clinical decision tools, or autonomous robotics — Pi uses the same Bearer auth, async job lifecycle, and typed response envelopes throughout.

Quick Start

Make your first API call in minutes — curl, TypeScript, or Python.

Authentication

Get your API key and learn how to authenticate every request.

API Reference

Full endpoint reference with request and response examples.

TypeScript SDK

Install @pi-api/sdk and start calling APIs with type safety.

​
What you can build

Brand Intelligence

Extract brand DNA from any URL and project it into structured context for your app.

Visual Commerce

Generate campaign-grade static ads and avatars with AI in a single API call.

Voice Agents

Configure reusable voice agents and start real-time LiveKit sessions.

Health & Neuro

Point-of-care triage, clinical decision support, EEG decoding, and wellness APIs.

Surveillance

Programmable video intelligence with real-time SSE incident streams.

Robotics

Agentic robotics runs with behavior orchestration and live event streams.

​
How Pi works

1

Get your API key

Obtain a Bearer token (e.g. pi_live_...) from your account. All requests use Authorization: Bearer <key>.
2

Submit a job

POST to any creation endpoint. Pi returns 202 Accepted with a job_id immediately — no blocking.
3

Poll for completion

Call GET /api/v1/jobs/:id with wait_for_completion=true to long-poll until status is completed or failed.
4

Use the result

Read the typed payload from data.job_result or retrieve the resource directly by ID.
Pi follows OpenAI-compatible conventions — Bearer auth, consistent JSON envelopes, and async job patterns — so it integrates naturally alongside existing AI tooling.