More than a workflow builder

11 node types get a workflow built. Browser automation, a data store and hosted models come with the platform rather than being wired up. 32 permission scopes, 37 audit event types, automatic dashboards and versioned deployments are what keep it running once real traffic and a real team depend on it.

Capability overview for a visual, event-driven workflow automation platform: AI agents with scoped tool access, MCP tool providers, built-in browser automation and a data store, role-based access control, permission-scoped API keys, an audit trail, monitoring dashboards and versioned deployments.

Least privilege, all the way down

32 permission scopes, API keys limited to a subset of them, an encrypted secret vault, and rate limits you set yourself.

Every change has a name attached

37 event types covering edits, deployments, approvals, permissions and membership — with the actor, the timestamp and the payload.

The dashboard builds itself

Throughput, errors and latency per workspace and per workflow, with the failing node named for you and no dashboard to build.

Ship a workflow like you ship software

Reviewable changesets, traffic weighted between versions from 0 to 100%, and configuration promoted between environments.

A step that decides what to do next

An agent node that calls real tools, on models included with your plan, with every reasoning step and tool result recorded.

Someone else's tools, on your canvas

A real browser, as a step in your workflow

A real browser as a canvas component — runs JavaScript, returns rendered content or a screenshot, with a per-workspace allowance checked before each request.

State that outlives the run

Streams, key/value records and agent memory as canvas components — scoped to your workspace, with retention you set.

Files and blobs your workflows can keep

Per-workspace object and file storage, scoped by a key derived from your workspace, reachable from a workflow and browsable from a console.