Learning by Reflection and Constraints

Igor bought a Fujifilm camera. That's not the interesting part. The interesting part is what he built to help himself actually learn photography instead of letting the camera collect dust in a drawer after two months. In this episode, we explore what Igor calls the "AI companion." Not an agent that does things for you, but a system that helps you stay with the things you actually want to do. The distinction matters more than you'd think. While the AI industry obsesses over agents that book your flights and manage your calendar, there's a quieter revolution happening. People are building personalized systems that protect them from distraction and help them follow through on their own commitments.Here's the pattern we're noticing: We're surrounded by too much choice. Every product, every service, every platform has adopted the same playbook. "It's up to you!" Maximum optionality. Complete personalization. The result? Paralysis. Netflix users browse for 45 minutes and watch nothing. People buy cameras, then abandon them when the learning curve meets the cognitive load of daily decision-making.The companion approach flips the script. Instead of delegating tasks to a machine because they're "beneath you," you're delegating a piece of your own agency to help yourself stay focused on what actually matters. Igor's camera companion knows his manual, his lenses, and his goals. It patiently guides him through the technical details without judgment, without pushing too hard, and is available whenever he has a question about aperture or wants feedback on a photo.What makes this moment interesting:- The effort required to build these personalized systems has collapsed. What once took weeks of automation setup now takes a conversation and a markdown file.- These companions are shareable. Just text files you can adapt to your own context. Think of them as recipes for commitment devices.- The emergence of "companion thinking" suggests something deeper about what we actually want from AI: not replacement, but augmentation. Not efficiency, but staying power.- And here's the uncomfortable question: Are we solving a systemic problem with individual solutions? Or building the tools we need to navigate a world designed to distract us?The conversation covers various topics, including photography as practice (referencing our episode with Christy George), the unbundling of coaching, and the reasons why typical AI demos fail to align with how humans prefer to make decisions. "Book me the cheapest flight!" sounds efficient. But that's not how we function. Chapter Markers 00:00 – Introduction: Agents vs. Companions 00:48 – Igor Got a Camera: Why Photography, Why Now? 05:49 – The Fujifilm Choice and Escaping the Phone 07:46 – Building the Companion: How It Works 10:14 – Agents vs. Companions: The Core Distinction 14:07 – Learning by Reflection: What the Companion Actually Does 18:46 – Why Does This Work? Automation vs. Augmentation 23:39 – The Choice Overload Problem 31:24 – Sharing Companions: Markdown Files as Recipes 43:25 – How to Get Started

---------------

You can also watch this episode on Youtube⁠

Follow the Rabbit feels like eavesdropping on a fascinating conversation between two well-read friends at a Berlin coffee shop—smart without being pretentious, critical without being cynical, and deeply engaged with contemporary culture while maintaining historical perspective. The podcast occupies a unique space between trend forecasting, cultural criticism, and philosophical inquiry, delivered with warmth, humor, and genuine enthusiasm for understanding how the world works.

Follow the Rabbit is hosted by Igor Schwarzmann & Johannes Kleske

Find out more about ⁠⁠Igor Schwarzmann⁠⁠ 

Find out moire about ⁠⁠Johannes Kleske⁠⁠