icCube 10, planned for release in November 2026, builds AI into icCube, and that AI works with each user's own access rights: if a user cannot see a dashboard, a schema, certain dimensions or a given cell, neither can the AI. Data governance stays in place: the AI only works with data each user is allowed to see.
The release will include a secure MCP server for AI agents, an AI chatbot (Olie) to talk with a schema, an AI Assistant that creates dashboards and widgets on demand, and an AI Assistant for the MDX console.
Why it matters: AI that sees only what you can see
The hard part of putting AI on business data is not the answer, it is making sure the AI never reaches data the person asking is not allowed to see. In icCube 10, the MCP server will have the same access rights as the user: an AI agent working for a sales manager in France will see what that manager sees, and nothing more.
icCube already secures data with dynamic MDX security, down to cell level. icCube 10 is designed so the AI works within those rights.
In icCube 10, AI will also be 100% opt-in. The AI features run on a separate, optional AI Server, and icCube works exactly as before if you don’t enable it. Existing and new roles get no AI features until they are granted through AI permissions.
What's coming in icCube 10
icCube 10 will add four AI features, all working with the user's own access rights.
- Secure MCP server: Lets AI agents query icCube with the same access rights as the user, so an agent cannot see a dashboard, schema, dimension or cell the user cannot see
- AI chatbot app (Olie): Lets users talk with a schema in plain language and get answers as tables, dataset summaries or manager-style reports, based only on the data they can access; it can also be opened inside your own application
- AI Assistant for Dashboard Editor: Creates dashboards and widgets on demand, writes and fixes MDX queries and calculated measures, arranges the layout, and guides you through the editor
- AI Assistant for the MDX console: Writes MDX queries in the MDX IDE, finds syntax and logic errors, suggests faster queries, and helps you explore your schema
What about the former AI Chatbot Widget? It is still there. This chatbot widget can now be linked to several widgets instead of a single one, so users can ask questions in plain language about the data all those widgets show.
Also on the roadmap: a dashboard assistant for viewers, an assistant for administration and schema creation, and an MCP server for scheduling and managing schemas.
How AI agents will work with icCube
In our July post on agentic workflows, we described the tools our agents use through MCP:
- MDX query validation: check a query before it runs.
- MDX query execution: run the query and return the result.
- Schema discovery: find the cubes, dimensions and measures to query.
- Schema validation in the Builder and dashboard validation in the Dashboard editor.
- Human in the loop: the AI does the heavy lifting, and your team validates the result.
What you can do today
You can prepare now, before icCube 10 is released: if you have a valid icCube license, you can already test a development build of icCube 10 on Docker Hub. It is not the official release, so contact us for directions before you start.
FAQ
When will icCube 10 be released? The official release is planned for November 2026.
Can I try icCube 10 before the release? Yes, if you have a valid icCube license: a development build is available on Docker Hub. Contact us for directions before you start.
Will the AI see data I am not allowed to see? No. The MCP server will have the same access rights as the user: if you cannot see a dashboard, a schema, certain dimensions or a given cell, neither can the agent using it.
Which AI assistants will be able to connect? AI clients that support the Model Context Protocol (MCP) will be able to connect to icCube 10's MCP server.
Will the AI features cost extra? No. The AI features are included in your icCube license at no extra charge. The only AI cost is the usage of the LLM itself, which stays on your side and under your control.

