Roadmap

Investing in the core of AI Analyst for better reliability, and for improved enterprise readiness.

Define row-level access rules in the semantic layer so users only see the data they're authorized to query.

Let enterprises bring their own LLM (e.g. Azure OpenAI, AWS Bedrock) so AI Analyst runs entirely within their own infrastructure and data governance boundaries.

Automated benchmarking framework to track AI Analyst quality and catch regressions on every change.

Allow users to interact with AI Analysts via Google Workspace Chat (1:1 and space-based conversations)

Connect external MCP tools (Slack, GitHub, Linear, etc.) to the Steward agent to extend its capabilities.