Managed agent runtime with durable sessions and tools for building applications
Let AI agents work in your signed-in Chrome browser with scoped access, human approval, and verified results
Stage
Not given
Not given
Founded
Not given
Not given
Based in
Not given
Not given
Pricing model
Freemium
Not given
Pricing
Personal: Free tier with daily free usage and pay-as-you-go credits. Pro: $20/mo. Team: $20/seat/mo. Enterprise: Custom pricing. Compute: $0.10/hr while active, $0 when idle.
Free: €0/month with 100 task credits per month and 1 connected browser. Pro: €5/month with 2,000 task credits per billing month and 5 connected browsers. Extra credits: 1,000 credits for €3 on Pro (do not expire).
Key features
Managed agent runtime with durable sessions
Dynamic workflows with model-generated JavaScript
Execution tracing through Langfuse
Web search capability
Decision tool with choice, score and questions
MCP and HTTP integrations
Scheduled agents with recurring or one-time execution
Platform connections for existing accounts
Line-item task receipts showing all costs
Support for multiple model providers
Redacted sensitive values including passwords, OTPs, card data, and API keys
Effect verification to prevent blindly repeating uncertain browser side effects
Integration with ChatGPT and Claude via MCP connectors
REST API and custom agent key support for own scripts
Task credit model with no separate browser-hour or AI model billing
What makes it different
Transparent per-task pricing with detailed itemized receipts showing every dollar
Flexible model payment: bring your own key, subscription, or pay at-cost rates
Usage-based billing separate from subscription plans
Free tier with no credit card required
OpenAI-compatible API using official client
Keeps your signed-in sessions in your own Chrome profile instead of copying them to a remote browser fleet
Brings the agent into your existing browser instead of moving your browser to a remote server
Charges only for accepted tasks, not for browser hours or AI model calls
Prevents uncertain side effects from being blindly replayed through durable task state and effect verification
Treats all page content as untrusted data so websites cannot grant themselves new permissions