Tasks & Approvals

Create tasks for your AI team, track progress, and review their work before it goes live.

Creating Tasks#

Describe what you need in plain English. The platform analyzes your request and routes it to the best agent based on skills and availability. You can also assign tasks directly to a specific agent, set a priority, and add context.

Task Lifecycle#

Every task follows a simple path: it's created and queued, then an agent picks it up and it moves to in progress. If the agent is supervised, the result goes to your approval queue as needs approval. Once you approve (or if the agent is autonomous), the task is completed and the output is available. If something goes wrong, it shows as failed and you can retry or reassign it.

Priorities#

Tasks can be assigned one of four priority levels:

• Low: Background work, no rush
• Normal: Standard priority (default)
• High: Should be handled soon
• Urgent: Needs immediate attention, jumps the queue

Search, Filter & Sort#

Find any task with full-text search across titles and descriptions. Filter by status, agent, department, or priority. Sort by date created, date updated, or priority.

Task Output#

When a task completes, you get:

• Result text: The agent's primary output (research findings, drafted content, analysis)
• Artifacts: Any files, documents, or structured data produced
• Duration: How long the task took
• Agent notes: Observations or follow-up suggestions from the agent

Reviewing Agent Work#

When a supervised agent finishes a task, it lands in your approval queue. For each item you can:

• Approve: Accept the work. The task moves to completed.
• Reject: Decline it. The agent gets your feedback and can retry.
• Request Changes: Send specific instructions. The agent revises and resubmits.

Bulk Actions#

Select multiple approvals with checkboxes and approve or reject them all at once. Useful when you have a batch of similar items, like a week's worth of social media drafts.