Databricks’ Genie App for Slack Is Now Generally Available

Databricks announced on November 12, 2026 that its Genie app for Slack is now generally available, putting permission-aware data questions, file analysis, charts, scheduled analyses and approved tool actions inside employees’ conversations.

Databricks describes the release as another front end for its existing governance model. Employees can pursue business context without leaving Slack, while administrators retain control over data retrieval, external actions and the audience for each answer.

Inside Slack, a worker can send Genie a direct message or mention @Genie in a channel or group DM. According to Databricks, the reply stays in the relevant conversation or thread, cites its sources and lets the user open the exchange in Databricks for more context.

Databricks makes one distinction explicit: user permissions govern what Genie may retrieve, while collaboration settings govern who may read the answer.

What can employees ask Genie inside Slack?

Genie answers questions using the requester’s existing access rights. Databricks says connected enterprise knowledge sources can include Google Drive, Confluence and Jira, while users can also upload files for analysis.

The app can return documents such as PDFs. It also renders tables and charts directly in Slack, keeping the initial result inside the conversation rather than sending the user elsewhere for every output.

One restriction remains. Databricks product documentation states that "charts do not appear when response visibility is private." The answer can remain private, but the chart presentation is unavailable under that setting.

Genie One is the default scope. It routes requests across the connected data estate rather than confining them to one selected information source or agent.

A conversation or channel can instead be pinned to a curated Genie Agent. Databricks says answers in that space remain within the agent’s data and instructions, giving administrators a way to create narrower areas for particular subjects or teams.

Channel owners can select the connected Databricks workspace. They can also choose public or private response visibility and determine whether the conversation uses Genie One or a pinned Genie Agent.

Recurring analyses can be created in natural language. Users can also use a scheduling button beneath an answer, with results delivered to the requester’s Genie DM.

In one exchange, an employee can ask a question across governed sources, receive a cited table or chart, and schedule the same analysis for later delivery without switching applications.

Can Genie take actions as well as answer questions?

Through MCP connectors to external tools, Genie can create or modify documents, comment on tickets and update records. Databricks says it uses the requesting user’s credentials for those operations.

The app prepares each change for approval before execution. This creates a visible checkpoint between a generated instruction and a change to a shared document, ticket or record.

Only the person who asked the question can approve or deny a tool action. When Databricks offers a scope selector, the requester has three approval choices:

• Apply approval once.
• Apply it always within that chat.
• Apply it always.

That approval model keeps the requester’s identity attached to an external change. It also gives administrators an auditable boundary between answering a question and modifying a connected system.

How do data access and response visibility differ?

Genie evaluates every request under the asker’s identity. Unity Catalog and connected-source permissions limit the information it can retrieve, according to Databricks.

Those retrieval permissions do not automatically limit the Slack audience. A public response, the default setting, appears to every member of a channel or group DM, including people who cannot access the underlying data themselves.

Databricks separates audience controls from source access. Visibility can be set for an individual, a channel or an entire workspace, and the settings follow a defined order:

• A personal choice applies to that user’s response.
• A channel-wide setting applies to the channel.
• A workspace setting applies across the workspace and takes precedence.

Workspace settings override channel-wide settings, which override personal choices. Databricks also lets administrators require every response to remain private, regardless of a user’s preference.

A response can therefore expose information to a wider Slack audience than the data behind it. The assistant retrieves material under the requester’s rights, but collaboration settings determine how far that answer travels.

This split makes audience visibility a separate administrative task from data governance. Teams must decide both who can query which records and who can read the resulting analysis.

Where does Genie process and store conversations?

Question processing and answer generation occur in the customer’s Databricks workspace. Databricks says conversation history is stored in that workspace’s region, keeping the retained exchange inside the customer’s selected deployment geography.

The central service that relays messages with Slack does not persist or log request or answer content. According to Databricks, it stores only the mapping between the Slack and Databricks workspaces.

The workspace’s geography and its cross-geography processing setting determine where large language model processing takes place. Those controls give enterprise teams separate visibility into storage and model-processing locations.

Deployment teams therefore receive two different location controls: the workspace region governs stored conversation history, while geography and cross-geography settings determine where the model processes requests and generates answers.

What deployment controls and limits ship with the app?

Unity Gateway provides centralized MCP policies, cost controls and monitoring. Databricks says it can produce traces of Genie’s actions and identify the users who requested them.

Audit logs identify Slack as the request origin. They record that source separately from activity originating through the Databricks web interface and Microsoft Teams, allowing administrators to trace actions by entry point.

Installation requires a Slack workspace administrator. After that administrator installs the app, users authenticate to Databricks and select a workspace.

Workspaces using Private Link need additional setup. A Databricks account administrator must configure a context-based ingress rule for the app before installation can proceed in that environment.

The compute model is automatic. Genie selects a SQL warehouse that the requesting user is authorized to access, but it does not offer the user a choice of compute.

Users also cannot create Genie agents from Slack. A specific agent can be pinned to a conversation or channel, but Databricks says that choice affects only new conversations.

Accuracy remains the user’s responsibility. Databricks warns that its agents "can generate inaccurate responses" and tells users to verify output before acting on it.

These limits make the release less configurable than a full data-analysis client inside Slack. Administrators control deployment, compute selection and agent scope, while users retain control over shared workspaces, source access and action approval.

Why does the Genie release matter for enterprises?

Databricks says the app turns Slack into another front end for its governance model. Employees can query business data and take approved action without leaving their conversations, while administrators keep control over retrieval, external changes and response audiences.

The product compresses the distance between analysis and execution. A cited answer can lead to a document update, a ticket comment or a scheduled analysis within the same working thread.

That convenience comes with a governance trade-off already identified by Databricks. Public answers can reach Slack members who lack access to the source records, even when Genie itself retrieved those records correctly under the requester’s identity.

For enterprise buyers, the release will succeed only if administrators can enforce clear audience limits while employees retain a direct path from cited answers to approved actions inside Slack.