Heapspace Meetup #39
Details
Join us on September 15 at 7 PM at the Startit (Savska 5) for two talks about AI and the ways it is changing our work. Oleg Ekimov will share his thoughts on how AI is changing the day-to-day work of business analysts and would like to open up a discussion with the audience on what has actually changed and where AI genuinely helps. Marko Nakić will talk about building and extending your own LLM agent harness.
≫≫ Please register here ≪≪
Doors open at 6.30 Talk starts at 7 PM.
*Please note that this event is in English!
***
Being a Business Analyst in the AI Era
speaker: Oleg Ekimov, Lead Business Analyst @EPAM Systems
AI is changing many aspects of a business analyst's work, but how much, and in what way, isn't always clear. In this session, I'd like to share my own observations and open the floor to your thoughts about what's actually changed in day-to-day BA work, what hasn't, and where AI genuinely helps vs. where it's more hype than substance.
----------
From LLM to Agent: Why You Should Build Your Own Harness
speaker: Marko Nakić, Junior DevOps Engineer @KubeCraft
An LLM becomes an agent only when a harness gives it context, tools, memory, an execution loop, and boundaries. In this talk, I will use pi, a deliberately minimal and highly extensible agent harness, to show why that surrounding layer matters as much as the model itself.
I will demonstrate how I extended pi with capabilities that matched my workflow, including more advanced subagent orchestration and features I wanted before similar options became common in other coding agents.
A small, programmable core lets you adopt the best ideas from Claude Code, Codex, OpenCode, and other harnesses without carrying every feature all the time: you add what you need, when you need it.
This approach requires maintenance and occasional debugging, but it also gives you control, helps you understand how agents work, and makes tinkering part of the fun.
Marko Nakić is a DevOps engineer focused on Kubernetes, GitOps, and automation. He uses AI tools and coding agents as his primary interface for everything, from writing code to many of the other tasks that are part of everyday computer work.