“Make no mistakes!” — Agentic engineering in 2026
Format: Workshop
Presenters: Lukas Trippe, Lukas Franken
AI tooling moves fast. New models, agents, and frameworks appear (and disappear) every week. In this session, we want to separate signal from noise and take a practical look together at where “agentic engineering” will stand in September 2026.
We’ll start with a quick lay of the land: Which models are out there? Which are open source? How do you run them, what do they cost, and how do you avoid getting locked into a single vendor? And crucially, how can LLMs unlock energy models for you?
We’ll climb the agentic engineering stack step by step: beginning where most of us already are (chatting with an AI assistant). We’ll look at how better memory and context handling improve your setup, and peek into the deep end: persistent agents that live on a server, remember what they’ve learned, and perform tasks for you even in the absence of direct control.
This is an interactive session. We’ll bring short inputs and demos, but want to open the room for discussion. We want to learn from each other’s experiences and tackle the questions that really matter: reproducibility, catching plausible but wrong results, cost, and the security and privacy risks of handing agents real permissions.
Curious? Skeptical? Both? Then join us.