We are entering one of the most exciting times in the history of software development.
AI is deeply embedded in our processes, expanding speed and scale, giving developers unprecedented freedom
—even as it transforms who and what software replaces?
AI doesn’t need to be perfect to replace every day execution work. It only needs to be fast and cheap. Learn how to stay relevant by building self-controlled agent systems instead of being replaced by them.
AI has moved way beyond chat interfaces and `copilot`. We are now entering a phase where models can plan, execute, evaluate, and iterate across real systems. Even learn from themselves and colleagues. Agent-based automation is already replacing large parts of computer-based work, not because the models are perfect, but because they are fast, cheap (due to crazy competition), and standardised into everyday workflows. This trend will not slow down. The real question for current developers is how to adapt and stay useful in the next 2–3 years.
This session is a practical, developer-oriented exploration of where we stand today and what that means for the near future and focuses on what is already happening in production systems. The central problem is not really `hallucinations` or lack of ethics, if any. The deeper issue is that modern AI is not “designed” at all. There are no intrinsic rules, no intent, and no built-in understanding of consequences. What we call intelligence emerges from probability, scale, and massive amounts of compute. Once such systems are connected to tools, APIs, and infrastructure, control becomes an architectural challenge rather than philosophical.
This talk introduces `self-controlled agents`: AI systems that are powerful, but embedded in boundaries that keep them comprehensible and governable. Using orchestration platforms such as n8n and Activepieces, combined with Drupal as a stable control surface.
The new Model Context Protocol (MCP) plays a key role here, so the agent (AI) knows what tool to use for best results into auditable building blocks. Its like an API wrapper that explains the skill to the tool.
Human judgment feeds this loop, not as a bottleneck, but as the final sense-check for actions that matter.
Rather than framing AI as the end of creativity, this session argues the opposite: developers are currently producing more creative output than ever. The risk is that creativity without structure becomes cheap, unvalued and thus replaceable.
Developers who understand how to combine AI, workflows, and human oversight will remain essential. Those who ignore this shift risk becoming obsolete, not because they lack skill, but because the execution layer has changed and got smarter by the speed of Moores law..
This talk is a warning, but also an invitation. AI is unstoppable, however we can still have a lot of fun in the next 3 years as developers.
This presentation is in English.
No specific technical knowledge required
Objectives:
- use AI to your benefit,
- use MCP so tools know what skills to use.
- We go a bit into philosophy and the future.
