The 30th Belgium NLP Meetup brought the community together at Ravical's offices in Ghent for an evening of conversations about what comes next in AI. Nearly a decade after the first meetup asked whether there was a market for NLP, this edition focused on a different question: how do we build AI systems that can operate effectively in the complexity of real-world work? Attendees heard from Kasper Rutten, Ben Schrauwen and Nico Gelders, who each explored a different aspect of the future of AI systems, from context-aware agents to autonomous applications and production-grade AI architectures.
Nicolas Gelders, Software Engineer at Ravical
Ben Schrauwen, Repeat founder
Kasper Rutten, Co-founder, CTO at Archer & Summit
One of the evening's recurring themes was that intelligence alone is not enough. In his talk, Kasper Rutten explored why professional work depends so heavily on context: prior conversations, organisational memory, domain expertise and countless implicit constraints that are difficult to capture in a prompt. Drawing on both software engineering and neuroscience, he argued that the future of AI agents depends less on increasing raw intelligence and more on improving how systems acquire, retain and act upon context over time.
The second half of the evening focused on what these ideas mean in practice. Ben Schrauwen presented Hermit, an experimental open-source project exploring autonomous applications where agents don't simply assist users but actively operate and evolve software systems. The evening concluded with Nico Gelders from Ravical, who shared lessons from deploying natural-language-to-code systems in production. His presentation demonstrated how safe code execution can help AI products move beyond traditional function calling and deliver more reliable, reproducible results when working with complex data and workflows.
