Building a Local AI Agent for Social Media: Demo & Q&A
Details
Ever wondered what it actually takes to build an AI agent - not a tutorial version, a real one you use every day? That's what this session is about.
You'll see a local AI agent that reads a LinkedIn feed every morning, decides what's worth engaging with, and drafts comments in a specific voice. You'll understand how it was built, what decisions went into it, and what didn't work the first time.
# What you'll walk away with
- A clear picture of how AI agents work in practice
- An honest look at the privacy decisions behind the architecture
- A realistic sense of what building something like this actually takes
# Format
Slides and a live demo, followed by Q&A. Bring your questions - but prepare to maybe not get an answer, as we are no experts either :)
# Who should come
If you work in UX or product design and you're curious about where AI is actually useful — not in theory, in your daily work — this session is for you.
# About Dunja
Dunja Cvetkovic is a UX & AI Designer with over 10 years of experience across B2B SaaS, biotech, fintech, eCommerce, and social impact. She started in graphic design and moved upstream into strategy and systems - which is a polite way of saying she got tired of designing things that shouldn't have been built in the first place.
She works at the intersection of strategic UX and ethical AI, which mostly means asking uncomfortable questions before anyone opens Figma. She runs structured experiments on AI tools, documents what actually works, and builds things herself before recommending them to anyone else. The agent you'll see in this session is one of those things.
Based in Germany, originally from Serbia, currently trying to automate everything that doesn't require genuine human judgment - and figure out where that line actually is.