AI business intelligence on governed data

AI business intelligence is business intelligence where an AI agent does the analysis: people ask questions in plain language, and the agent finds the right data and metrics, runs the queries and explains the result, instead of an analyst building every report by hand. The answers are only useful if they are consistent and correctly permissioned, so in Databrain the agent answers from a semantic model your team defines, within the permissions of the person asking, with row and column security, roles and SSO on the Business plan. Dashboards and scheduled reports run on the same definitions, and builders can reach them from Claude Code or Cursor through the MCP server.

Yes. Connect a warehouse or database, define your metrics once in the semantic model, and your teams use them through dashboards, the Databrain agent and scheduled reports. Builders can reach the same model from Claude Code or Cursor through the MCP server. If you later want analytics inside your product, the Enterprise plan puts the same layer there.

Yes, and for one person exploring their own data, that can be enough. Databrain is for when the answers reach other people. It keeps one set of definitions for every agent, answers each person within their own row and column security, roles and SSO, runs health checks across the pipeline, and can put the same governed data in front of your customers. Your builders keep their own tools: the MCP server lets Claude Code, Cursor and your own agents query it.

No. Databrain reads the same warehouse as Power BI, Tableau or Looker, so the two can run side by side. You can start with the questions your dashboards do not answer, through the agent, and move dashboards over later onto the same connection and metric definitions.

Yes. Every plan includes model access from Databrain, and every plan can also connect the model provider your security team has approved, such as OpenAI, Azure OpenAI, Anthropic or Gemini. If you bring your own model, your provider bills the model usage.

Databrain uses role-based access control: workspace members are Admins, Editors or Viewers, and the Business and Enterprise plans add configurable roles and permissions, row and column security and single sign-on. The Databrain agent answers each person within that same access. Shared dashboard links carry an expiry time you choose, and administrators can review activity in the audit logs, which record authentication events and administrative actions.

Yes. Databrain's report scheduler emails selected charts from a dashboard on the recurring frequency you choose, using the email settings configured for your workspace. On Solo, reports and alerts go to you; from the Team plan, they go to the teammates and stakeholders you list, so people who only need the numbers get them without logging in.

Adding viewers does not, because viewers are free on every Databrain plan. Internal plans are priced per builder seat plus usage. Solo is $39 per month with one builder seat, Team is $199 per month with three (more at $75 each, up to ten), and Business is $3,125 per month with ten builder seats, SSO and roles. Agent work uses credits: each question in the app, call through the MCP server or API, or scheduled run uses 1. Solo includes 500 a month, Team 3,000 and Business 25,000, and work beyond that is metered. Reading an existing dashboard or result uses none. Paying annually takes 20% off Solo, Team and Business, and every plan starts with a 14-day free trial.