Agentic Harness - AI Meetup
Two hands-on builders, four short talks and one open Q&A panel on what it really takes to make AI agents work: from getting more out of Claude Code to shipping production systems that save money and keep users engaged.
🎤 Meet the speakers
Roland Szabo is an AI consultant who has been doing AI and ML before it was cool. He likes to explore new ways to work with LLMs, to augment our minds.
Filip Tasadan is an AI engineer who builds production LLM systems, from document extraction pipelines to chatbots that let customers talk to their own data. He focuses on the practical side: making AI products cheaper to run and more useful to the people who use them.
🧠 What they’ll talk about
Session #1 (Roland): Advanced features in Claude Code. A deep dive into the Claude Code features most people never touch, and how to use them to speed up real development work.
Session #2 (Filip): Reducing cost on invoice extraction. How to cut the cost of running LLM-based invoice extraction without losing accuracy.
Session #3 (Roland): BYOH, Build Your Own Harness. What an agentic harness actually is, what goes into one, and how to build your own instead of relying only on off-the-shelf tools.
Session #4 (Filip): Lessons learned from building a production chatbot to talk with your own data. What worked, what broke and what drives engagement once real users show up.
🗓️ Agenda
18:00 Doors open & welcome
18:15 Roland: Advanced features in Claude Code
18:35 Filip: Reducing cost on invoice extraction
18:55 Roland: BYOH, Build Your Own Harness
19:15 Filip: Lessons learned from a production chatbot
19:35 Q&A panel with Roland & Filip
20:00 Networking
🚀 Join us
Whether you’re already building with agents or just curious about where to start, come for the talks and stay for the conversations. Spots at the Cowork are limited, so register now to save your seat, and bring a friend who’s been talking about AI agents nonstop!