# "OpenFlow: run AI agents as a workflow"

<dl class="home-points">
  <div><dt>Draw the workflow</dt><dd>Each step is an agent with a prompt, a model, and tools. Connect steps to say which results feed which.</dd></div>
  <div><dt>Run steps in parallel</dt><dd>Steps that don't depend on each other run at the same time. A step starts when everything it needs is ready.</dd></div>
  <div><dt>Stay in control</dt><dd>Open any agent's conversation to answer its questions, approve a tool call, or retry a failed step.</dd></div>
  <div><dt>Bring your own model</dt><dd>Use OpenAI, Anthropic, ChatGPT sign-in, Amazon Bedrock, OpenRouter, or a local model such as Ollama.</dd></div>
</dl>

## How a run goes

1. **Draw the workflow.** Add an agent for each step. Give it a prompt, a model, and tools, then connect it to the steps that feed it.

   [![The canvas with a feature brief splitting into risk review, test strategy, and implementation, then joining at a delivery plan, with the Inspector open on the brief agent.](https://philbotar.github.io/OpenFlow/media/screenshots/offline-search-authoring-inspector.png)](https://philbotar.github.io/OpenFlow/media/screenshots/offline-search-authoring-inspector.png)

2. **Run it and watch.** Steps that don't depend on each other run at the same time. Open any agent's conversation to answer a question, approve a tool call, or retry a failed step.

   [![A finished run: the planner and checklist agents are done, and the checklist agent's conversation shows the handoff it read and the result.](https://philbotar.github.io/OpenFlow/media/screenshots/release-checklist-run.png)](https://philbotar.github.io/OpenFlow/media/screenshots/release-checklist-run.png)

3. **Review what happened.** History keeps every run. Open a replay to read each agent's conversation and the handoffs between them.

   [![The History tab listing a completed run of the release checklist workflow.](https://philbotar.github.io/OpenFlow/media/screenshots/release-checklist-history.png)](https://philbotar.github.io/OpenFlow/media/screenshots/release-checklist-history.png)

## Get started

1. [Install OpenFlow](https://philbotar.github.io/OpenFlow/getting-started/). The macOS app needs no development tools.
2. [Connect a provider](https://philbotar.github.io/OpenFlow/guides/provider-setup.html) so agents can call a model.
3. [Build your first workflow](https://philbotar.github.io/OpenFlow/guides/first-workflow.html), or [start a direct chat](https://philbotar.github.io/OpenFlow/guides/direct-chats.html) if you only want to talk to one agent.

## Find what you need

| I want to… | Read |
| --- | --- |
| Understand how it works | [How OpenFlow works](https://philbotar.github.io/OpenFlow/concepts/how-openflow-works.html) |
| Design a workflow | [Workflow authoring](https://philbotar.github.io/OpenFlow/guides/workflow-authoring.html) |
| Handle approvals, retries, and failures | [Runs and recovery](https://philbotar.github.io/OpenFlow/guides/runs.html) |
| Look up a setting | [Configuration](https://philbotar.github.io/OpenFlow/reference/configuration.html) |
| Fix a problem | [Troubleshooting](https://philbotar.github.io/OpenFlow/troubleshooting/) |
| Build it from source | [Build from source](https://philbotar.github.io/OpenFlow/getting-started/build-from-source.html) |
| Browse everything | [All documentation](https://philbotar.github.io/OpenFlow/overview.html) |