Tasks
Tasks turns recommendations, Agent output, and team intent into executable work. Instead of jumping between tools, Tasks gives you a single queue for what needs to happen next.
Open Automations from the sidebar (Tasks appears as Automations in the navigation).
What Tasks are for
- Turn an approved recommendation into executable work
- Track operational to-dos across your marketing data stack
- Attach workflows (pipeline runs, syncs, transformation jobs) to a task
- Create tasks manually or via the Agent and MCP
Journeys and pipeline orchestration are accessed through Tasks — there is no separate Workflows nav item.
Task views
Tasks supports two views:
| View | Best for |
|---|---|
| List | Scanning all tasks with status filters |
| Kanban | Visual workflow by status column |
Filter by status to focus on what needs attention: pending, in progress, completed, or blocked.
Task detail
Each task includes:
- Description — what needs to happen and why
- Status — current progress
- Attachments — supporting context (charts, documents, data references)
- History — audit trail of status changes and actions
- Workflow steps — attached pipeline or job execution
- AI suggestions — optional next-step recommendations from the Agent
Pipeline canvas for kind='pipeline' tasks opens directly from the task detail page.
Create a task
From the UI
- Open Automations and click New Task.
- Add a title, description, and optional due date.
- Attach a workflow if needed.
- Save — the task appears in your queue.
From the Agent
Ask the Agent to create a task as part of a conversation:
“Create a task to investigate the Meta ROAS drop and attach the attribution transformation run.”
From MCP
Use the tasks_create MCP tool to create a task and optionally attach an underlying workflow in one call. See MCP Server for setup.
Related
- Approvals — where AI recommendations enter the review queue
- Pipelines — operational graph behind task workflows
- Agent — create tasks from conversation
- MCP Server — programmatic task creation