> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.runtype.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.runtype.com/_mcp/server.

# What is Runtype?

Runtype is an AI product development platform. Use it to build, test, deploy, and monitor the Flows and Agents behind AI features in user-facing products.

## Who is Runtype for?

Use Runtype if you build an AI-powered experience for a software company, product team,
startup, or agency. Runtype hosts your Flows and Agents, versions them, and serves them
through Surfaces, so you don't host a backend for them yourself.

If your agent already runs on your own infrastructure, you can keep it there and connect
it to Runtype to use Logs, traces, and eval capture with it. For more information, see
[Bring your own agent](/developer-guides/guides/bring-your-own-agent).

## How Runtype works

Build a Flow or Agent, attach it to a Product as a Capability, and expose the Product
through one or more Surfaces. Use Records for structured context and Tools for actions or
external data. Test configurations in Evals or the Playground before you deploy them.

The following table shows how the main platform objects fit together:

| Object  | Role                                          | Connection                                |
| ------- | --------------------------------------------- | ----------------------------------------- |
| Flow    | Runs a repeatable sequence of steps.          | Add it to a Product as a Capability.      |
| Agent   | Lets a model select Tools and sequence calls. | Add it to a Product as a Capability.      |
| Product | Groups related Capabilities.                  | Add one or more Surfaces.                 |
| Surface | Routes requests to a Product's Capabilities.  | Exposes the Product through an interface. |

Use these concepts throughout the platform:

* [Flows](/user-guide/flows/what-are-flows): Run a fixed sequence of prompts, API calls,
  transforms, and conditions when you can define the steps and branches before each run.
* [Agents](/user-guide/agents/what-are-agents): Let a model select Tools and sequence
  calls toward a goal.
* [Skills](/user-guide/agents/agent-skills): Package instructions that an Agent loads only
  when a task needs them.
* [Products](/user-guide/products-surfaces/what-are-products): Group related Flows and
  Agents as Capabilities, then deploy them through one or more Surfaces.
* [Capabilities](/user-guide/products-surfaces/adding-capabilities-to-a-product): Attach a
  Flow or Agent to a Product so that the Product's Surfaces can route requests to it.
* [Surfaces](/user-guide/products-surfaces/what-are-surfaces): Expose a Product through
  channels such as a chat widget, a REST API, an MCP server, A2A, Slack, and email.
* [Records](/user-guide/records/what-are-records): Store structured data with JSON
  metadata so that Flows and Agents can use it as context.
* [Tools](/user-guide/tools/what-are-tools): Add actions that Flows and Agents use to
  search the web, call APIs, run code, or connect to external services.
* [Evals](/user-guide/evals/what-are-evals): Compare model and prompt configurations for a
  Flow or Agent with the same test inputs, and review output, success rate, latency, cost,
  and token usage.
* [Schedules](/user-guide/schedules/what-are-schedules): Run a Flow or Agent at a specific
  time or on a recurring cadence.
* [Logs](/user-guide/logs/what-are-logs): Review Flow and Agent executions, inspect
  failures, and find the step to fix.

Use the [Dashboard](/user-guide/dashboard/what-is-the-dashboard) to review account
activity and open the main product areas. Use the
[Playground](/user-guide/playground/what-is-the-playground) to test prompts, models,
settings, and Tools before you save a Flow or Agent. To build from your editor or AI
assistant instead, click **Connect your assistant** on the Dashboard to connect the
[Runtype MCP server](/developer-guides/integrations/runtype-mcp-server). To call Runtype
from your own code through the API, SDK, or CLI, see the
[Developer Guides](/developer-guides/getting-started/introduction).

## Example: deploy a support Flow

Alex works on a support team at Example Organization. They create a Flow that drafts
responses to customer questions, attach it to a Product as a Capability, and add a
REST API Surface. Their application sends requests to the Surface, which routes each
request to the Capability. Alex uses Records for support context and Evals to compare
model and prompt configurations before deployment.

## Platform keys and bring your own key (BYOK)

Start with [platform keys](/user-guide/getting-started/platform-keys-vs-byok) to use
popular AI models without creating provider accounts or adding provider configuration.
To use your own provider credentials, [connect AI model providers](/user-guide/integrations/connecting-ai-model-providers)
in Runtype.

BYOK is available on the Team plan and higher. On those plans, you can use platform keys
and BYOK at the same time. Use your own keys when you need direct provider billing,
dedicated rate limits, or access to more models.

## Next steps

Continue with these guides:

* [Creating your account](/user-guide/getting-started/creating-your-account): sign up and confirm that you can open the Dashboard.
* [Understanding the Runtype UI](/user-guide/getting-started/understanding-the-runtype-ui): find your way around the Dashboard before you build.
* [Quickstart: build a social media post generator as a REST API](/user-guide/getting-started/quickstart-social-media-post-generator): build a Flow and expose it through a REST API Surface.
* [Quickstart: From Agent to chat widget](/user-guide/getting-started/quickstart-from-agent-to-chat-widget): build an Agent and embed it as a chat widget.