Last week I had the chance to present at SLO AI Lightning Talks #2, a meetup bringing together folks interested in AI from across Slovenia.

My talk explored AI, used in customer support & the human need for... well, humans (or human-like responses).

As companies rush to deploy chatbots and automated support systems, customers increasingly find themselves trapped in loops, desperately searching for that "talk to a human" option. What if we gave our AI chatbots the tools to respond in a human-like manner?

The presentation covered:

- Why "efficiency" isn't always what customers want
- When AI helps vs. when it frustrates
- Finding the right balance

On the technical side, I touched on our stack at Clover Labs - we've been experimenting with DSPy (via Ax LLM, a TypeScript implementation) for building more reliable AI pipelines. It's a different approach from prompt engineering - you define signatures and let the framework optimize the prompts for you. Early days, but promising results so far.

It was great to connect with others thinking about these questions. The AI community in Slovenia is small but growing, and events like this are exactly what we need to push the conversation forward.

Thanks to the organizers for having me!

---

P.S.: See the (not so aesthetic) presentation of the talk below!

Hi, I'm Blendor

Posts