From dashboards to decisions: how Fluid transformed Thermatic’s business data with AI

Thermatic wanted to introduce a chatbot to its Microsoft Power BI dashboard to allow users to interrogate data and receive meaningful, natural language answers without requiring advanced reporting knowledge. 

The goal was to make it easier for users across the business to access key information quickly, reduce the reliance on manual report analysis, and encourage greater use of data when making operational decisions.   

In addition to improving accessibility, Thermatic wanted to explore how AI could support greater adoption of reporting tools by making data analysis more approachable for users with varying levels of technical expertise.  

Fluid worked closely with the Thermatic team to ensure the AI solution interpreted business data accurately and provided relevant responses based on the existing dataset. 

A multi-step approach was used to review and clean the data model, adding messages and instructions for the agent, addressing ambiguities such as date, time and any industry-specific language. This integrated with Microsoft Fabric (an AI-powered data platform designed to simplify data management and analytics), which brings together data engineering, storage and reporting in one connected environment – with Power BI sitting on top as the reporting layer, drawing on the same underlying data foundation. 

 This meant Fluid could work from a single, consistent data source rather than juggling disconnected systems, and conducted extensive testing with both internal and client-generated questions. 

The project involved assessing and refining the data model to improve accuracy and consistency, training the chatbot to understand industry-specific terminology and business context, and creating instructions and prompts to guide responses and improve reliability.  

Finally, Fluid conducted extensive testing to a wide range of real-world questions, paying particular attention to ensuring that the chatbot would be able to understand different ways users may ask questions, such as languages, abbreviations and operational terminology.  

The result was a chatbot capable of providing meaningful answers based on Thermatic’s data, creating a more intuitive and accessible reporting experience for users.  

Early testing demonstrated the potential for AI-powered reporting to improve how users interact with business data.  

By allowing users to ask questions in their natural language, the solution reduced the need to navigate multiple reports and dashboards, helping teams access information more quickly and efficiently.  

The project highlighted the importance of ensuring that data quality and business context are considered during AI implementations. The more effectively the underlying data is structured and maintained, the more valuable the resulting insights become.  

There were a few challenges during testing; Microsoft Fabric’s capacity limitations

occasionally resulted in service interruptions when large volumes of questions were submitted, requiring systems restarts during testing. While this added complexity to the project and made the testing process more difficult than initially anticipated, it also helped identify capacity requirements and performance considerations for future deployments. 

The testing process provided valuable insights into how users interact with data through conversational AI highlighted opportunities for further optimisation. As client testing continues, the solution is expected to further improve accessibility to business intelligence and support faster, more informed decision-making across the organisation.  

Speaking of the project, Krystal Simpson, the developer at Fluid who worked on the project, said: “Our goal was to make Thermatic’s reporting data more accessible to a wider range of users by enabling natural language interaction with Power BI, built on the data foundation we set up in Microsoft Fabric”.  

“To achieve this, we worked closely with the Thermatic team to refine the data model, provide the chatbot with the necessary business context, and ensure it could accurately interpret industry-specific terminology. Extensive testing helped us continually improve the quality and reliability of responses, creating a solution that demonstrates the potential of AI to make business intelligence more intuitive and accessible.” 

 Andrew McKay at Thermatic, added: “We wanted to make it easier for our teams to access key business information without needing advanced reporting expertise. Working with Fluid has allowed us to use AI to help simplify data analysis and improve engagement with our reporting tools.  

“The chatbot has already shown promising results during testing, enabling users to retrieve insights more quickly through natural language queries and helping us build a more data-driven approach to decision-making.”  

If you’re looking for support for an upcoming AI project, get in touch with our team here.