Your growth workspace. Your choice of AI.
Use the Overview agent to work with your connected workspace in conversation, or connect a compatible external AI through MCP. Research, creative preparation, and publishing work share the same account context rather than living in separate, disconnected chats.
How the workflow works
- 01
Choose your interface
Open the Overview agent, or configure a compatible MCP assistant with authorized workspace access.
- 02
Give a specific task
Include the intended outcome, channel, required assets, and boundaries. Resolve missing information instead of accepting invented inputs.
- 03
Inspect the action
Review the draft or job, approve required production steps, and confirm the resulting library or publishing state.
Ask for an outcome with enough context to act
Give the agent a concrete task: inspect an SEO and GEO finding, prepare a post, create a video draft, or schedule supported content. Include the intended audience, destination, and constraints that matter. The agent can use connected data and Brand Identity rather than asking you to paste your company explanation every time.
When a required asset, connection, or choice is missing, the task needs clarification. Supplying an approved portrait for a talking video or naming the publishing account is better than letting the model guess. Check the proposed output and action status, not just the conversational response.
Create and schedule through the same workspace
The agent can prepare supported content and creative workflows, including video planning with the available production options. A request for AI motion should remain an explicit production choice, not an assumption that every video consumes motion resources.
Review drafts, storyboard approvals, account selections, and schedules where the workflow requires them. A generated asset, an accepted background job, and a delivered post are different states. The workspace library and publishing queue let you inspect what actually happened.
Bring a compatible external assistant through MCP
Model Context Protocol exposes supported tools so an external assistant can work with workspace data and actions. It is not a copy of every dashboard screen and does not make every AI application compatible. Your assistant must support MCP and be configured with the required connection.
Tool availability and account scope determine what it can do. Keep credentials private, grant only appropriate access, and review the assistant's requests before consequential actions. The external assistant does not gain ownership of your workspace simply by connecting.
Turn evidence into a deliberate next action
An agent-ready improvement brief is useful because it includes concrete findings rather than a vague instruction to improve SEO. The assistant can inspect the affected pages, propose changes, and explain what needs review. Observed citations and search performance remain evidence, not promises of future results.
Use the agent for reducing handoffs: research a finding, prepare a draft, and move approved work toward a schedule. Human review remains valuable for product accuracy, recipient suitability, publishing permission, and claims about your business.
Example workflow
From a search finding to a scheduled explanation
A founder gives the assistant an evidence-backed brief about an unclear product comparison. The assistant prepares an explanation using approved facts, creates a social draft pointing to the relevant resource, and helps schedule it for a named account. Each step stays inspectable; the assistant does not silently change the website or claim a ranking win.
Common questions
Can any AI application connect?
Only applications that support MCP and the required connection can use the exposed tools. The built-in workspace agent is available directly in the dashboard.
Does MCP expose every dashboard feature?
No. It exposes supported tools and actions. The connection must have the right account scope, and some workflows still require review or preparation in the product.
Does a chat confirmation mean my post is published?
Not necessarily. Check the job and delivery state in the workspace. Draft creation, background production, scheduling, and final delivery are separate steps.