Overview

Build and run your own agent platform.

Every team building agents builds the same system from scratch:

• A server to run the agents (batch, streaming, or background mode).
• A database for sessions, runs, traces, and memory.
• Auth and RBAC, validated via JWT on every request.

This system is called an agent platform and the goal is to build the foundation once so we can build agents and workflows without worrying about the underlying plumbing.

We’ll build our agent platform using FastAPI and Postgres and it will be able to:

• Run agents in batch, streaming or background mode.
• Store what agents read and write into organized sessions, runs, traces and memory.
• Authenticate every request using JWT tokens, providing necessary security and permission guarantees.

Coding Agents Love Agent Platforms

An agent platform ties the system together, so every new agent slots into the same runtime, storage, and connectors.

Coding agents love this proximity. Everything lives in one place: agent code, logs, traces, evals, the live service. Claude Code can manage the entire agent development lifecycle. From creating an agent, to improving it, extending it and evaluating it.

Step-by-Step Guide

Here’s the step-by-step guide to build an agent platform managed entirely by coding agents.

Step	What happens
Run Locally	Run your agent platform (AgentOS + Postgres) locally using docker.
Create an Agent	Use Claude Code to create a new agent.
Improve an Agent	Use Claude Code to read container logs and improve an agent.
Run on Railway	Run your agent platform on Railway.
Run Evals	Lock in behavior with regression tests.
Next Steps	Teams, workflows, scheduling, and Slack interfaces.