99.2% on WebVoyagerState-of-the-art browser and computer-use agents
Developers

Run digital workersfrom your own code.

Launch any saved workflow from your own systems over the API, connect coding agents through MCP, and read the results back as structured data.

Read the docsBook a demo
curl https://api.ramain.ai/api/v1/automations/$WORKFLOW_ID/runs \
  -H "Authorization: Bearer $RAMAIN_API_KEY" \
  -H "Idempotency-Key: rate-sheets-2026-10-12" \
  -H "Content-Type: application/json" \
  -d '{"inputs": {"investor": "FLC", "date": "2026-10-12"}}'

# 202 Accepted: a run_id, plus links to the
# run's status, result and events
Quick start

From a task to an API call.In three steps.

  1. 01

    Build and save the workflow.

    Describe the task in the Workbench and watch a digital worker run it in a live cloud browser. When the run succeeds, save the workflow and enable it.

  2. 02

    Copy its input schema.

    API Docs in the app shows each workflow's generated schema and examples. Names, dates and files become typed inputs.

  3. 03

    Launch it from your server.

    Send one authenticated request with an idempotency key. The API accepts it with a run ID you follow until the result is ready.

Workflow API

Five endpoints.One run to follow.

Paths are relative to https://api.ramain.ai/api/v1. Every request carries a scoped key in the Authorization header.

POST /automations/{workflowId}/runs
Launch an enabled workflow with its inputs. Needs the workflows:run scope.
GET /runs/{runId}
Read the run's status and links. Needs runs:read, as do the reads below.
GET /runs/{runId}/result
Read the outputs once the run ends, or its pending status until then.
GET /runs/{runId}/events
Read a snapshot of the run's events, step by step.
GET /runs/{runId}/recording
Download the browser session video when it is available.
MCP

Give your coding agenta digital workforce.

The Ramain MCP server connects Claude, Cursor and other MCP clients to your workspace. In Claude, add it as a custom connector and sign in. In Cursor or a script, use a personal access token.

Build workflows.
Start an authoring session, guide the builder and save the result without leaving your agent.
Run and watch them.
Launch runs, follow them step by step and read the outputs when they finish.
Hand back to a person.
When a run needs a human, your agent sees why. The decision stays with someone on your team.
{
  "mcpServers": {
    "ramain": {
      "url": "https://api.ramain.ai/mcp",
      "headers": {
        "Authorization": "Bearer ${env:RAMAIN_MCP_TOKEN}"
      }
    }
  }
}
Run updates

Know when the work is done.Or when it needs you.

Follow a run from your own system, or have Ramain call you back.

Callback URL
Add a public HTTPS receiver to the launch request. Ramain calls it with the run, including the metadata object you sent.
Run events
Read the run's event snapshot to show progress inside your own product.
Human checkpoints
When a run waits for a person, its status carries the instructions and a live view link. Your system answers through the checkpoint endpoint.
Safe retries
If a launch response is lost, resend the same request with the same idempotency key. You get the original run back instead of new work.
Security

Your keys start runs.Never your credentials.

01

Scoped API keys.

Each key carries only the permissions it needs, such as workflows:run to launch and runs:read to follow. Revoke a key at any time.

02

Credentials stay in the vault.

Passwords, keys and one-time codes are used at run time. They never pass through your API calls or the AI model.

03

Keys stay on your server.

Send keys in the Authorization header from server-side code, never in a URL or in browser JavaScript.

04

Every run on the record.

Runs started from code appear in run history with their inputs, events and session recording.

Resources

Everything you needto build on Ramain.

Guide

API quickstart

Save a workflow and launch it from your server, step by step.

Read the guide →Reference

API reference

Endpoints, the run envelope, files, checkpoints and errors.

Open the reference →SDK

Browser SDK

Embed a live browser view of a run in your own product.

Read the SDK docs →MCP

MCP server

Connect Claude, Cursor and any other MCP client to your workspace.

Set it up →Systems

Connectors

Every system and channel a digital worker can work in.

Browse connectors →Docs

Product docs

Workflows, agents, triggers, the inbox and workspace security.

Open the docs →

Ship your first workflow.We build it with you.

Our team maps one process with you and gets it running from your own code.

Book a demoRead the docs