Most companies sit on years of hard-earned knowledge — but have no way to turn it into AI that’s private, reliable, and actually useful.

In this episode, I’m joined by João Carvalho, a machine learning PhD who spent years at ETH Zurich studying why AI fails, and Wael Abdelmalek, former AWS architect who helped global enterprises turn their expertise into production-ready AI systems.

They met at a hackathon in Zurich and quickly realized they were solving the same problem from opposite sides. That moment became Uthereal — one of Switzerland’s top startups and a platform that lets companies build secure, business-grade AI systems trained entirely on their own data, with their own guardrails, and without exposing anything to Big Tech.

We talk about what it really takes to build trustworthy AI, why so many GenAI pilots never make it to production, and how companies can protect their data while still moving fast.

What You’ll Learn

- Why AI Fails: A clear look at hallucinations, reliability gaps, and why “80% accuracy” isn’tenough for expert domains.

- Building Private AI: How Uthereal creates business-grade “ChatGPTs” that stay inside your cloud — fully private, fully controlled, and built on your own knowledge base.

- From Pilot to Production: Why 95% of GenAI projects stall after the demo, and what companies need to do differently to make AI work in the real world.

- Security & Trust: What most teams don’t realize about data leakage, model backdoors, and the risks of sending sensitive information to public LLMs.

The Future of Work: Why AI is an empowerment tool, not a replacement — and where human judgment still matters.

Uthereal’s story is a reminder that real AI value isn’t about chasing hype — it’s about turning what your business already knows into something scalable, secure, and built to last.

Learn More

Website: uthereal.ai/

LinkedIn: João

LinkedIn: Wael

Uncomplicated: Humans vs. Machines is produced by Urban Podcasts.