Scottish Water Tames Data With AI Chat

Scottish Water has turned its complex capital investment data into a conversational asset, ditching cumbersome reports for direct, natural-language queries. By implementing Databricks Genie, the utility company has empowered its project teams to get instant answers about project status, financials, and risks, all within their daily workflow.

For years, Scottish Water’s Capital Investment (CI) program generated vast amounts of data, but accessing it was a significant hurdle. Information was siloed across numerous reports, often requiring a data specialist to extract the relevant figures. This created delays, duplicated efforts, and meant that crucial project insights didn't always reach the decision-makers who needed them most. StartupHub.ai data shows that Databricks, a leader in this space, holds a strong score of 82/100, indicating its significant influence in the data and AI platform market, competing with giants like Palantir Technologies (85/100) and Alphabet (79/100).

From Report Hunting to Direct Questions

The solution, internally branded as SPARK, integrates directly into Microsoft Teams via Copilot. Users type a question in plain English, and a supervisor agent orchestrates the query. This agent connects to a Databricks Genie Space, which translates the question into a query against governed data residing in Unity Catalog. The result is then returned to the user in seconds. This drastically cuts down the time spent searching for information; a process that once involved multiple clicks and dashboard loads can now start with a single question.

Practical questions like "List all open project risks expiring in August" or "What is the current live risk score for project X?" are now met with immediate, reliable answers. This shift means less time spent locating static reports and more time acting on dynamic insights.

Faster Access, Less Friction, Broader Use

The impact on efficiency is substantial. What might have taken 4-5 steps to find in disparate reports now becomes a direct question-and-answer exchange. If 100 users ask just three questions weekly, it could save between 520 to 1,300 hours annually. Beyond speed, the system provides answers tailored to specific inquiries, reducing the need for users to interpret broad-audience reports. It also lessens the reliance on IT or data specialists, making governed insights accessible to a wider audience.

Allan Mason, Programme and Project Delivery Manager at Scottish Water, commented, "SPARK is going to completely change how our portfolio and project teams interact with data. It moves us from static reports to real-time, intelligent conversations with our information, empowering our people to make quicker, better-informed decisions and unlocking value we simply couldn't reach before."

Built for Trust and Scale

Crucially, the system is designed for trust. Governance is embedded from the start, with Genie operating on data within Unity Catalog and utilizing shared semantic definitions. This ensures consistency and explainability. Scottish Water curated its data into a "gold layer" and built a semantic layer with metric views to standardize business terminology, minimizing ambiguity. The Genie space was further tuned with specific business rules, terminology, and user phrasing to ensure accuracy and relevance.

Ongoing monitoring tracks adoption, answer quality, and performance. This includes analyzing conversation duration, recurring questions, and query execution times. The solution is also packaged using Databricks Asset Bundles for repeatable, consistent deployment across development, testing, and production environments, managed through Azure DevOps.

This implementation of Databricks Genie demonstrates a powerful path for organizations to democratize access to trusted business data. By combining well-governed data with a conversational interface, teams can work more efficiently and make better-informed decisions, moving beyond the limitations of traditional reporting.