How to build startup-level AI architectures that actually scale — without a data science army.

In 2025, building AI systems isn’t just for Big Tech anymore. With today’s tools and frameworks, small teams — or even solo founders — can create production-grade AI systems. The real challenge is not building something that works, but building something that keeps working.
Drawing from my experience leading enterprise-scale operations and now building AI-first infrastructures at the Swiss AI Chatbot Factory, I’ll share a practical blueprint for creating reliable, maintainable AI architectures that scale with minimal human overhead.
This talk focuses on the architectural thinking behind sustainable AI startups: modular design, automation chains, agent orchestration, and fallback logic. You’ll see how to avoid the common traps of “prompt spaghetti,” vendor lock-in, and untraceable LLM behavior — and instead design systems that are transparent, testable, and business-aligned.