How Crawbl Works
Overview
Platform flow
A person asks
A user submits a request.
Crawbl coordinates
The platform receives the request and determines what work needs to happen.
Agent work happens in isolation
AI work executes in isolated runtimes.
Isolation enables scale and control.
Results come back clearly
Results return as answers, suggestions, or approval requests.
Users see outcomes, not implementation details.
- A normal chatbot answers once. Crawbl keeps working after the first answer.
- One request can turn into follow-up actions, approval checks, saved context, and more work in the background.
- That is the difference between a tool that talks and a tool that helps finish something.
Detailed view
Client interfaces
Users interact through product surfaces such as the app, API, or realtime connections.
This is where requests begin and results are delivered back.
Orchestration layer
The control plane handles routing, access checks, approvals, provider mediation, and integration boundaries.
This is where the platform decides what should happen next.
Runtime layer
Agent work runs in isolated runtimes instead of directly inside the client-facing system.
This separation gives Crawbl stronger control, scale, and safety.
- The app only needs to collect the request and show the result.
- The platform handles the harder parts underneath: routing work, checking approvals, and keeping state.
- This keeps the product simpler and makes the platform easier to grow over time.
This page is the public view. For the deeper component-level explanation, see the system overview.