At the beginning of April 2026, Daniël Spee and I presented at Voxxed Days Amsterdam. We explained the resemblance between a good whisky and a good AI Agent. After workshops and presentations on RAG and Agents, the focus this time is on creating and evaluating high-quality agents. It becomes practical with a demo using Spring AI, OpenAI, Dokimos and Langfuse.
The presentation is brand new. The reviews were ok. There were around 150 people in the room. The rating for the talk is close to 4 out of 5. I was satisfied, but it could be better. Luckily, the talk is recorded. I do not like watching myself on these videos, but I always watch the recordings to learn. One piece of feedback was about looking at the audience. At specific moments, it feels more like I am talking to Daniël than to the audience.
After the talk, someone from the audience approached us and complimented us on the talk. He thought the content was way ahead of a lot of the other talks. We did not provide only the shiny demo; we also showed how to improve your applications and make them ready for production through evaluations. That is, of course, the drive to keep doing these kinds of talks.
I’d like to hear what you think about the talk. If you want to talk about evaluations or have us come to your conference, drop me a message.