Sapiom

One API key gives agents access to 80+ tools: web search, deep search, browser automation, screenshots, 400+ LLM models, image generation, text-to-speech, sound effects, and phone verification. Pay-per-use with spend governance built in.

View on sapiom.ai

Documentation

Introduction

Sapiom gives agents cloud capabilities. Call those capabilities directly, or use them inside agents you can test locally, deploy, schedule, and operate.

What do you want to do?

Build an agent Start in Agent Studio, or connect Sapiom to Claude Code or Codex. Build and test locally before you deploy.

Use a cloud capability Search, browse, run compute, store data, or use another Sapiom capability through Cloud MCP or application code.

Operate deployed agents Open the Agents dashboard to inspect agents, builds, production runs, alerts, and settings.

Build an agent

Agent Studio is the integrated path. Install the desktop app or launch the same workspace with npx, then choose Claude Code or Codex when you start a coding session. Studio manages the project tools and gives you Canvas, local testing, Deploy, and Prod Run controls.

If you prefer your coding agent’s own terminal or editor experience, connect Sapiom Project MCP directly instead.

Install Agent Studio Choose the desktop app or the npx host, then start with Claude Code or Codex.

Connect your coding agent Add Sapiom Project MCP to Claude Code or Codex without opening Agent Studio.

Authentication at a glance

You can create, edit, check, and run an agent locally without a Sapiom account. Sign in when you link or deploy a project, start a production run, manage cloud triggers, or call a cloud capability.

Your coding-agent account is separate from your Sapiom account. Claude Code uses your Anthropic authentication; Codex uses your OpenAI authentication. Sapiom browser sign-in authorizes project-to-cloud actions, while standalone Cloud MCP and SDK connections authenticate independently.

Continue to Build your first agent. You will create the shipped starter and complete a genuine local run before any cloud action or capability spend.

Use a capability directly

You do not need to author an agent for a one-off cloud operation. Connect Sapiom Cloud MCP to a supported client, or use the @sapiom/tools package from application code. See Ways to use Sapiom when you want the full surface map.