Has self-service BI finally arrived with AI?
I came across this article by Silvan Späti that explores how conversational AI, combined with the Model Context Protocol (MCP), might finally deliver on the long-standing promise of self-service Business Intelligence (BI).
The value for the end customer
The idea is powerful: instead of relying on developers or data analysts to build reports, business users could simply ask questions in natural language. For instance, an operations manager could ask, “Which of our wind turbines are underperforming against their contractual power curve for the last quarter?” and receive an immediate, data-backed answer. This direct interaction has the potential to dramatically speed up decision-making and empower teams to explore data on their own terms and time.
In my experience, end-users often consume data from a dashboard only to re-process it. They combine several numbers or present the information in a different way for their specific needs. This is where an LLM-based approach could be particularly beneficial, making it much easier to fine-tune queries and get the desired end result without manual work or overloading the dashboarding or development team.
A dose of healthy scepticism
However, I must admit I am still a bit sceptical. Will end-users truly adopt this workflow, or will they just keep asking the development team for help? The shift from being a passive consumer of dashboards to an active conversational partner with data is a significant behavioural change that should not be underestimated.
The real win
That said, the real value might lie in two key areas. First, the ability to communicate in natural language is a fundamental shift. It lowers the technical barrier significantly.
Second, and perhaps more importantly, is the ability to verify the output. The article explains how the system can show its work, including the queries it runs. This means you can fact-check the AI’s conclusions yourself, ensuring it is not hallucinating and that the insights are grounded in actual data. This builds trust, which is essential for adoption.
It makes data more accessible, results more tailored, and more verifiable. The combination of human expertise guiding a powerful, transparent AI tool could be where the true revolution lies.
