Flanders AI Forum 2026
20/10/2026
The Flanders AI Forum 2026 brought together researchers, companies and organisations from across Flanders to explore developments in artificial intelligence and their practical applications. The forum featured presentations, discussions and demonstrations, as well as opportunities to connect with the Flemish AI ecosystem.
Thank you to everyone who joined us. With more than 1,500 attendees, the Flanders AI Forum 2026 was a great success.

FAIR at the Flanders AI Forum
How do you turn groundbreaking AI research into concrete business value?
At the Flanders AI Forum, FAIR researchers and partners showed how AI research is translated into practical tools, methods and applications for Flemish organizations.
FAIR Deep Dives
13:15 - 14:20
Hoe taalmodellen de Vlaamse context begrijpen en waarom dat uw klantinteractie, marketing en compliance raakt.
Why does this matter for customer interaction, marketing, HR and compliance?
Researchers and industry partners discuss FlemBench, the VLAM programme and culturally grounded AI systems. They show why language, context and local knowledge are essential for AI applications in Flemish organisations.
With: Veronique Hoste (UGent), Walter Daelemans (UAntwerpen), Luk Overmeire (VRT), Tanguy Coenen (imec) and Jelle Demanet (ENT.A).
@ The Impact
13:15 - 14:20
Duurzamer produceren met AI-agenten: wat werkt écht op de werkvloer, en hoe zet u de eerste stap?
What really works on the shop floor, and how do you take the first step?
Discover how AI agents can support sustainability goals through environmental monitoring, concrete business applications and direct access to research knowledge.
With: Shane Ó Seasnáin (VITO), Stijn Vranckx (VITO), Carlo Massimo Manna (VITO) and Lennert De Smet (KU Leuven).
@ The Pulse
15:00 - 16:05
Ontdek hoe betrouwbare generatieve AI de industrie transformeert.
Discover how reliable generative AI can support industry.
Explore practical methods to improve the reliability, robustness, safety and explainability of generative AI systems through industrial use cases and downloadable starter kits.
With: Bart Van Doninck (Flanders Make) and Mihail Mihaylov (Sirris).
@ The Pulse
Demo's
NextGen Food Inspection: Teaching AI the Defects We Rarely See - Flanders Make and ILVO
Edge AI: An Easy and Intuitive Way to Deploy Your AI Models - Flanders Make
Met AI naar veiligere steden: slimme inzichten voor onze leefomgeving - VITO
Kunnen we de AI-agent vertrouwen? AI-veiligheid in real-time - Ghent University
Chat with an Adhesive Expert: AI-Powered Selection - Flanders Make
Optimizing Process Operation Despite Variable Usage - Flanders Make
Talk to It, Show It, Watch Your Robot Work
Organisation: Flanders Make
Two novel teaching paradigms for robot programming will be demonstrated in which no custom coding is needed:
Tell Your Robot: in this demo, you can speak to a robot and it will understand which object it should pick and where it should place it. This is a form of ‘agentic robotics’, leaning on the power of large-language-models such as Chat-GPT.
Show Your Robot: in this demo you will teleoperate a robot to execute a given task. An AI is then trained on the combination of live images and robot controls. This allows for the learning of complex skills where visual feedback is important. A second robot will show a live demonstration of a pre-trained robotic skill (folding).
NextGen Food Inspection: Teaching AI the Defects We Rarely See
Organization: Flanders Make and ILVO
Generate realistic potato defects in real time to enrich AI training data with rare or difficult-to-capture examples. Automatically annotated images reduce manual effort and help build more robust defect detection and sorting systems.
Identify and Track Any Asset with VLMs
Organisation: KU Leuven
Use advances in Visual Language Models (VLMs) to detect and identify any asset in an unknown environment. Use this information to guide users towards certain objects, or create a semantic map to keep track of them.
Edge AI: An Easy and Intuitive Way to Deploy Your AI Models
Organisation: Flanders Make
Deploying AI models in low-resources processing units require a full set of steps to optimize the models and match the industrial requirements such as accuracy and real-time. The Edge Tower developed by Flanders Make and validating AI models (including the ones developed in FAIR) allows a fast and intuitive deployment of AI models in low-resources CPUs and GPUs. The demo will showcase some examples with an interactive interface to monitor performance of AI models.
Met AI naar veiligere steden: slimme inzichten voor onze leefomgeving
AI for Safer Cities: Smart Insights for Our Living Environment
Organisation: VITO
Discover how AI turns complex environmental simulations into fast, usable insights. Learn about our research on the living environment and test SAFE-CRANE: an AI tool, developed with the construction sector, that maps local wind risks around cranes and construction sites.
Van complexe planning naar slimme beslissingen
From Complex Planning to Smart Decisions
Organisation: KU Leuven
How can AI and optimization solve complex planning problems? We demonstrate applied research in which we combine exact optimization with machine learning, language models and interactive visualizations.
How Can We Put AI to Work in Agriculture?
Organisation: ILVO
AI technology is developing at lightning speed. With computer vision, our hoeing and spraying machines already work more precisely than ever. With time-series AI, harvest data can be determined, and AI is finding its way into farmers' decision-making software. There are already many autonomous machines, but robots do not yet manage the farm by themselves. There is still a long way to go, and important steps remain to be taken. Is this step possible thanks to recent developments in AI? Visit our stand and see for yourself how, within the Flanders AI Research Program, we are preparing our robots for this!
Can We Trust the AI Agent? AI Safety in Real Time
Organisation: Ghent University
Agents are not always trustworthy. Discover how we can check AI responses from AI agents in different domains before they reach users or the real world. Safe actions go through, while other cases are referred to a human or blocked. The system can take over and steer towards safe alternatives.
AI for Semantic Segmentation
Organization: Flanders Make
Digital twins of manufacturing environments enable realistic data generation for AI training, validation and production simulations. This demonstration combines robotic picking and assembly, autonomous navigation and scheduling to evaluate production plans and how they adapt to disruptions.
Chat with an Adhesive Expert: AI-Powered Selection
Organization: Flanders Make
A chat assistant combines large language models with a rule-based expert system built on real adhesive expertise. Ask a question in plain language and see how the system evaluates captured expert knowledge to provide a traceable recommendation for selecting adhesives and tests.
Optimizing Process Operation Despite Variable Usage
Organization: Flanders Make
Discover how processes can be optimized when usage conditions vary. The demonstration shows how adaptive control can improve efficiency compared with conservative setpoints, and how varying loads complicate condition monitoring and predictive maintenance.

