Hi, I'm
Ivan Lučin
VP of Engineering at Productive. Frontend engineer under the hood, now building AI agents. This is where I keep the random things I make and say — in English and Croatian.
Latest
- ToyWIP
Fleet
Supervise and steer every Claude Code session, on every machine, from one place — a Rust CLI, a phone-first web app and a macOS app over Tailscale.
English⌥ Source - Talk
Building an Agent Harness: Architecture & Patterns Deep Dive
How Productive rebuilt its AI agent harness three times for a vertical SaaS with 50,000 users — the architecture and patterns that survived.
EnglishInfobip Shift 2026Zadar, Croatia30 min▶ Video - Post
When Writing Code Is No Longer the Issue, What Is?
We now need to build habits, systems, and culture around AI coding — on slop, review as the new bottleneck, the validation loop and the AI vampire effect.
English1 min read↗ Inc. - Episode
Productive AI Podcast
A conversation about building Productive's AI platform and agents — what we're building, how, and what we're hiring for.
EnglishProductive Podcast1 h 18 min♫ Listen - Post
Testing the Test
Code in automated tests should also be tested — how to keep tests honest by minimising logic, sharing structure, breaking the code on purpose and measuring how good your tests really are.
English1 min read↗ Productive Engineering Blog - Post
AI Engineer's Challenges — Part 1: Intro
Prompts are code and the model is the runtime. Why AI-enhanced software is easy to get 50% working, hard to push to 80%, and impossible to finish.
English2 min read↗ LinkedIn
Browse
Writing
Writing on engineering, leadership, frontend and AI — in English and Croatian.
7 items →
Playground
Small hobby coding experiments — toys, games and tools I build for fun.
3 items →
Talks
Conference talks and meetups — slides and recordings.
3 items →
Podcasts
Podcast episodes I hosted or guested on.
3 items →
In the media
Interviews, features and mentions — articles about my work around the web.
5 items →