Engineering systems in the AI shift

Doors open

Registration, drinks & snacks

‍

Talk: AI-native architecture

Software engineering today requires AI-native architectural practices: evolvable contexts, structured collaboration between humans and AI agents, and governance that keeps systems reliable as they learn and change.

This session explores what happens to engineering systems when AI becomes part of the architecture itself.

We’ll look at what could be a natural next step: AI-native architecture as an engineering capability.

Coffee break

Catered food, drinks, networking

Talk: AI agents as operating systems

Modern agents increasingly resemble operating systems in how they coordinate context, tools, memory, and constraints.

Using a real production agent as an example, this session explores:

• structural parallels to operating systems
• what’s genuinely new (self-evolution, identity persistence, embedded ethical boundaries)
• how to reason about agents as infrastructure rather than features

A practical framework for teams building agentic systems beyond prototypes.

Coffee break

Catered food, drinks, networking

Roundtables: System-level AI adoption

Six practitioners will share short inputs, which will be followed by small-group discussions where participants exchange experience and compare approaches.
‍

1. Organic AI adoption: lessons from vibecoding

AI adoption doesn’t always come from top-down initiatives, it might grow organically through experimentation. Based on internal vibecoding sessions, Andrei will share what actually stuck and what failed.

👉 What patterns of AI usage have naturally emerged in your team?
‍

2. Knowledge debt in AI-native systems

When agents generate a growing share of code, teams may lose the ability to understand and explain their own systems.

👉 Should we treat this knowledge debt as a critical risk to manage, or accept it as a black box and compensate with stronger testing?
‍

3. AI as a structural stress test

Code generation became cheap, so the main bottleneck now is decision-making. Unclear priorities, fragmented context, and slow decisions are exposed under speed.

👉 Can your current decision-making model operate effectively in a world of fast and cheap execution?
‍

4. Aligning leadership and engineering in AI adoption

Executives see AI as a magic wand, while engineers see slop-generating crap. Both perspectives are valid, but this tension stalls adoption.

👉 How do you align expectations and operate effectively if you’re the tech manager caught in the middle?
‍

5. Centralizing AI adoption in skills, workflows, and reasoning
‍

The first step in AI adoption was teaching developers how to use new tools. The next step is more structural: centralizing AI-related skills, embedding AI into workflows, and making reasoning itself part of the process.

👉 How do you turn AI usage from an individual habit into a system-level capability? 

Networking & afterparty

More time to connect over food and drinks.

Talk to peers and get more clarity on how to design AI-native systems, how to adopt it in organizations, and learn how others are doing it.