I write about building AI products in production.

I'm Liam Tengelis. I build AI systems and write about what that actually looks like: the architecture decisions, the failure modes, and the distance between what works in a demo and what survives real customers.

I've built and shipped AI features that went into production at scale. I built a practical guide to language models for engineers working with LLMs. I've led work on deep research agents and AI-native product integrations - the kind of systems where the hard problems start after the initial demo lands well.

I started Black Tusk Data to write honestly about what it actually takes to build AI products in production. Not the demo version, but the one that has to survive customers.

The newsletter is called After the Demo. That's the space I care about: after the proof-of-concept, after the launch announcement, after the first customer complaints, when the real engineering work begins.

If that's useful to you, subscribe to After the Demo.