FIN Intelligence, designed with user-centric innovation, aims to bring building automation systems (BAS) to the next level of simplicity for end users, system operators, energy managers, and maintenance staff. By combining structured, tag-based data with the power of large language models (LLMs), FIN makes it possible to turn everyday questions into powerful, behind-the-scenes analysis.
Our Intelligent Assistant is designed to harness the transformative potential of generative AI to perform complex and labor-intensive tasks, then deliver the results in clear, conversational language. The use of generative AI makes FIN easier, but how? Let’s take a step-by-step look at energy usage in a building, following the data and the AI behind the scenes from start to finish.
For example, a future intelligent assistant would be prompted with a natural language question: "How is my energy consumption this week?" Behind the scenes, the AI would convert that prompt to something that is machine-readable, in the form of a Haystack query. The structured data for historical energy usage would be easy to identify using tags, starting with site, which identifies the building and the siteMeter tag combined with the date, which defines the timeframe, will gather last week's electricity usage. The AI would then compare mathematically the weekly energy usage for this week and report back in a natural language response the results. "Your building's energy consumption for the week is 3200 kWh, which is 12% lower than last week."
The user can then ask project-specific questions to drill down into additional insights: "Which areas of my building are consuming the most energy?"
The AI would take that prompt and convert it into a machine-readable query. The areas of the building are identified using the following tags: starting with the floor tag to identify the physical levels of the building, combined with the space tag to locate the areas. For the energy consumption, the query would look like combinations of elec and meter tags plus the submeterOf tag referencing the main site meter of the building. The AI would then compare mathematically the energy usage for each area and report back in natural language. "The server room accounted for 42%, office area 30%, and the lobby at 10% energy consumption."
From here, you can ask the assistant for additional advice: "Can you suggest ways to reduce energy in the server room?" The intelligent assistant then goes beyond the historical data for the site to access a larger body of knowledge (LLM). What's going on behind the scenes is that keywords, such as "server room" and "energy reduction," are passed to the AI, and the neuro network looks for weighted results to determine the best response for saving energy in a server room.
It then responds back with three suggestions: "Increase cooling set point, schedule regular maintenance, implement better airflow management."
What looks like a simple question and answer exchange is actually a sophisticated workflow where FIN Intelligence combines Haystack-tagged data, analysis, and generative AI to deliver actionable insights in seconds. Instead of manually crafting queries and digging through trend logs, users can rely on the Intelligent Assistant to translate natural language into precise, machine-readable requests and return clear insights and recommendations.
Learn more about FIN Intelligence's beta program and stay tuned for more behind-the-scenes examples on how AI can be utilized to improve the building automation.
Topics from this blog: Project Haystack OEM Smart Buildings Smart Equipment Energy management FIN Framework Technology Industry Building Automation System Single Pane of Glass AI FIN Intelligence GenAI
Back to all posts
