At the Peak of Data and AI 2025 conference, two standout sessions showcased how people are combining FME with AI to transform their operations in creative, scalable, and inspirational ways. Whether you missed the live sessions or just want the highlights, here’s a look at what these talks covered, and how to achieve incredible things with AI and FME.
Watch the recordings of both sessions in our encore webinar: Peak of Data & AI Encore: AI-Enhanced Workflows for the Real World
AI Meets FME: Smarter Workflows at Scale
In the first presentation, Oliver Morris of Avineon-Tensing took us through some amazing real-world applications of AI-enhanced workflows powered by FME. He highlighted how far AI has come in the past 18 months, with cheaper, faster multimodal models now capable of handling complex tasks like document classification, data extraction, and image analysis. AI isn’t just for data scientists or software engineers anymore: with platforms like RoboFlow, AWS Bedrock, and FME, non-technical users can now build no-code workflows to analyze images, classify data, and trigger automated actions.
In this powerful use case from UK Power Networks (UKPN), they leveraged FME and Google Gemini to extract information from over a million handwritten historical electrical service records. The task took just over a day with FME and AI, saving an unfathomable amount of manual effort and hundreds of thousands of pounds.
In this use case from Amberside Energy, the goal was to automate the classification and organization of large volumes of site imagery collected from solar and energy sites. With an FME pipeline connected to Bedrock and Anthropic models, those images are now automatically analyzed, organized, and summarized into reports. Issues are flagged in Microsoft Teams chats, turning forgotten folders into valuable operational insights.
If you’re working with unstructured data—messy images, cryptic shapefiles, or any other data type—the combination of FME and AI makes it possible to bring order to chaos, turning it into structured data accurately and at scale.
Oliver also shared Tensing Labs’ Shapefile Fixer tool, a web app that ingests poorly labeled shapefiles, processes them with FME, summarizes them using DuckDB, and then prompts Gemini to suggest more intuitive field names. What used to take hours of manual review can now be accomplished in minutes.
A great tip from Oliver when using AI for complex projects: “Break down the problem, and make each task so simple the AI can’t get it wrong.” When working with AI, this equates to tasks running quickly, cheaply, and at scale.
Your AI Assistant and Problem-Solver
The second session, presented by Dmitri from Safe Software, took the form of a lightning round of ideas. He showcased many ways to enhance workflows using AI. From generating textures for 3D models and optimizing batch scripts, to transforming user voice messages into professional QA reports in real time, Dmitri offered a whirlwind of inspiration.
For example, by embedding a QA agent in FME Realize, users can select a pipe feature and instantly get AI-driven feedback on its geometry and grade. In another example, a voice message describing a field issue was transcribed, translated, and converted into a formatted report within seconds.
The message: AI can be your assistant, your problem-solver, and your creative partner.
Key Takeaways
-
AI is Ready for the Real World: You don’t need to wait for better models. Today’s tools are already powerful, affordable, and accessible.
-
FME is the Glue: Its ability to orchestrate complex workflows, handle data preprocessing, and validate outputs makes it the ideal partner for AI.
-
No-Code Platforms Are a Game-Changer: Tools like RoboFlow make advanced tasks approachable for anyone—not just developers.
-
Microtasking is the New Best Practice: Break big problems into small, AI-friendly tasks to maximize accuracy and minimize risk.
-
Flexibility Is Built In: Thanks to FME’s robust connector ecosystem, you can use whatever AI model suits your needs, whether it’s Gemini, Claude, or any other model out there.
Try It
If you’re ready to start experimenting with AI in your own workflows, check out our FME and AI page for our learning resources. Browse FME Hub for all the AI connections you can start using today, like Gemini, Bedrock, OpenAI, and more.
Legacy Data is traditional and values the wisdom of the past. He’s the go-to for office lore and frequently reminisces about ‘the early days’. His fax number is still operational and his office shelves are filled with paper files, plus a floppy disk or two (or twenty). Find him by the water cooler solving a Rubik’s cube or recollecting on how things used to be done. Just don’t ask him to download your pdf or move to the cloud.