MCP
Connect external tool servers to DataGenie over the Model Context Protocol so Wisdom's agent can use them while it reasons - permissioned, governed, and scoped.
MCP (Model Context Protocol) lets you connect external tool servers to DataGenie so Wisdom's agent can use them while it reasons. A connection brings a set of tools into reach - for example, a Slack connection exposes tools to read a channel or post a message - and Wisdom's agent picks the right tool for the step it's on, mid-reasoning, then folds the result back into its answer.
MCP connections extend what Wisdom's agent can do without ever routing your governed data through them uncontrolled. Every tool call is permissioned, and read vs write is marked on each tool. See Responsible AI.
How to add a connection
Open MCP in the Knowledge Center
Navigate to Knowledge Center from the sidebar, then select MCP. This lists the tool servers already connected and their scope.
Click Add connection
Click Add connection and point DataGenie at the MCP server you want to connect - for example a Slack server. Provide its endpoint and any credentials the server requires to authenticate.
Review the tools it carries
Each connection carries a set of tools. Review them before saving - every tool is clearly marked read or write (for a Slack connection, reading a channel is a read tool; posting a message is a write tool), so you always know which tools can only observe and which can take action.
Choose a scope and save
Set the connection to Global or Yours (see below), then save. Wisdom's agent can now reach for those tools whenever they fit the question it's working through.
Start read-only. Connect the read tools you need first, confirm the agent uses them well, then add write tools once you're comfortable with how they behave.
Two scopes
Global
Shared with everyone on your DataGenie instance. Use Global for connections the whole team should be able to draw on - a shared Slack workspace, a common ticketing system. Governed centrally by an administrator.
Yours
Private to you and available only in your own conversations. Use Yours for connections tied to your personal access, so the tools appear when you ask Wisdom but not for anyone else.
How the agent uses connections
Wisdom's agent acts through your connections, never around your controls. It selects a tool only when it fits the step it's reasoning through, each call is permissioned, and write tools take action only within what the connection allows. Your data stays governed - no raw data is handed to the LLM, and read vs write stays enforced on every tool. See Responsible AI.
What's next
Documents
Upload your manuals, SOPs, and regulations to the Knowledge Center - DocuFeeder turns them into citable knowledge Wisdom answers from, with sources.
Memories
Wisdom remembers the durable things about how you analyze — the metrics you care about, how you like answers framed — and applies them to every answer. Confirmation-gated and private to you.