SSE Airtricity serves hundreds of thousands of customers in Ireland with their electricity and gas energy needs. With smart electricity meters becoming widespread, they wanted to offer improved energy insights with a solution they can continue to innovate on. The team built an in-house AI-powered recommender on the Databricks Data Intelligence Platform that delivers personalized energy-saving advice to over 65,000 smart meter customers. Using Databricks’ AI tooling — from foundation models to MLflow for governance and AI judges for quality control, they launched in just four months. With personalized energy usage insights flowing directly to consumers, they’ve experienced significantly greater customer engagement that’s driving measurable behavioral change toward their net zero goals.
Getting smarter about energy
Serving around 750,000 customers, SSE Airtricity is one of the three largest energy suppliers across the island of Ireland. As customer expectations continue to evolve, delivering intelligent, personalized experiences they’ve grown accustomed to from retail and media industries has become table stakes. With a clear commitment to leading the next wave of energy innovation, SSE Airtricity has focused on developing tools that transform how customers interact with and optimize their energy usage. SSE Airtricity had the data to support this shift, with smart meters deployed across a large portion of its customer base. Still, basic usage visualizations and generic tips like "switch to LED bulbs" fell short of expectations.
‘We were originally looking at black box solutions from third parties, but felt these offered limited ability to understand the customer engagement with specific features. Being able to refine and expand capabilities based on what customers valued was essential’ said Michael Mulloy, Head of Data and Analytics at SSE Airtricity.
The team realized they needed a fresh approach. They wanted to build something in-house that would give them complete control and flexibility to innovate quickly and deep visibility into customer engagement. But entering the world of generative AI for the first time for Airtricity's Data and Analytics team felt daunting. Luckily, Databricks was there to help.
Building a strategic advantage through AI
SSE Airtricity's Data & Analytics team decided to build the Enhanced Smart Insights (ESI) Energy Advisor entirely in-house on the Databricks Data Intelligence Platform. By leveraging smart meter data and customer survey responses, the ESI Energy Advisor delivers highly personalized, actionable advice designed to help users optimize their energy consumption. For instance, by identifying EV owners and guiding them toward off-peak charging schedules, the system transforms raw data into significant, actionable cost savings.
Having already migrated from Oracle to Databricks, the team was starting with a robust data platform, with Unity Catalog providing centralized governance for both data and AI models, and Delta Lake for reliable storage and a foundational data architecture for their AI strategy.
The challenge was ingesting billions of rows of data and 48 daily readings per smart meter to feed models and deliver actionable insights to customers. Lakeflow Spark Declarative Pipelines simplified the ingestion and transformation of this data to support their AI models, enabling Spark workflows to process billions of rows every night.
The ESI Energy Advisor is built on Anthropic's Claude Sonnet 4.5, accessed through Databricks AI Model Serving, allowing the team to deploy generative AI without managing infrastructure. MLflow became central to providing complete observability and traceability, ensuring optimal quality improvement.
The business acceptance breakthrough came from taking an evaluation-first approach: implementing AI "judges" using MLflow's GenAI evaluation framework alongside human feedback collection from the early stages and using custom quality metrics to establish alignment before writing production code. Through a unified interface, they can define evaluation criteria for models, agents and insights before they reach customers, checking for length, language, tone and accuracy. MLflow Traces provides visibility into model performance, while the fully customized judges automatically reject and regenerate any failing insights.
“The judges and feedback loop gave the business confidence that the solution was robustly governed,” said Maria Leacy, Analytics Manager at SSE Airtricity.
Custom monitoring apps built with Databricks Apps track costs, model performance, and AI judge failure rates in real time, delivering production-ready AI with thresholds and automated alerts.
Helping customers and helping the business
The ESI Energy Advisor was soft-launched in mid-December 2025 and exceeded expectations. "We went from proof of concept to production in about four months,” said Derek Aherne, Lead Data Scientist at SSE Airtricity. “The accelerators and code from Databricks really accelerated us forward.”
Within the first month, SSE Airtricity saw a 12% increase in web visits to the insights section compared to the previous month. The team anticipates that this will continue to grow over the coming months.
This shift from passive browsing to active engagement is an important first step in the customers’ net zero journey. By consuming actionable insights, customers are no longer just monitoring energy — they are actively optimizing it. 90% of logged-in customers now navigate to the insights section to engage with the recommendations, and 10% click to view their complete set of personalized insights. Customers are also saving their personalized insights for future reference, which not only enhances their overall experience but also opens opportunities for SSE Airtricity to recommend relevant energy services tailored to their needs, such as solar installations and low-carbon home upgrades. By focusing on delivering useful advice and solutions that genuinely benefit customers, these interactions foster trust and engagement, ultimately motivating the behaviors necessary to reach their net zero goals.
There were also notable operational benefits. The solution eliminated the need to manually create generic content and the complex business logic required to map insights to customer profiles. Insights now refresh automatically every two months when customers receive their bills, giving SSE Airtricity an advantage over competitors.
"A competitor requires customers to wait 120 days after signup to receive insights,” said Maria. “With us, the day after you get a bill, you can have your insights. It's a real differentiator.”
The Energy Advisor, grounded in data and AI, gives SSE Airtricity a substantial strategic advantage. The team can now rapidly adapt to market changes, whether it's upcoming dynamic pricing, new tariff structures, or emerging programs such as virtual power plants and flexibility schemes. With governed foundation models already in production, SSE Airtricity is now extending the Databricks Platform to power their next features: tariff recommendations, bill analyzer, peer comparisons and energy disaggregation — empowering the smart meter customers in Ireland with clearer, more personalized insights that help them use energy more efficiently and support a more sustainable energy future.