🥳🎂

I agree 100%. This is exactly the foundation upon which I've been building for the last two months 💪

"Don't spend months learning. Learn and then build." This sentence is really important. Of course, studying first is essential, but real knowledge comes from getting your hands dirty, trying things, and making mistakes. Stay curious 🫡

Been through exactly this, so a few things I'd tell my past self: Don't build the second brain as a separate knowledge base. The moment it lives outside the workspace it becomes a second source of truth, and the two start drifting apart at the first update. Nobody (human or AI) can tell anymore which version is current. What worked for me, in ICM terms: I stopped treating the workspace as "workflows + a wiki somewhere" and structured the root itself as an ICM artifact. The root is a container, not a pipeline: it holds L0 (identity and operating law), L1 (routing maps), L3 (stable distilled knowledge) and L4 (working material), and deliberately no L2, because the root itself runs no repeatable process. The full five-layer workspaces (your production stages and workflows) live inside it as self-contained processes. The container organizes and routes; the processes produce. Three rules that made it work: One canonical home per information type. Brand identity, offer, operational know-how: each lives in exactly one place. Processes link to it, never copy it. A copy is a second source of truth from the moment it exists. Keep L3 and L4 apart. What the AI must follow (distilled rules, references) is not what it must process (raw sources, drafts). Mixing them is what makes lookups unreliable. A routing map at the root (L1), so the AI navigates instead of searching. The map answers "where do I go"; the folders own the content. So: integrate, don't separate. The knowledge layer sits next to your workflow docs, not inside them and not in a parallel wiki. I'm deep into this exact problem right now: I'm building my own system around it, a base container any business could start from and grow inside. Happy to compare notes if useful.

@Mark G Good question. If it's not action-based yet, it's not a process. So it doesn't get a process shape either. It just needs a stable home. What's working for me: every topic gets a small folder. An L3 (references) for the distilled stuff the AI treats as ground truth. An L4 (source) for raw material, if there's any worth keeping. A routing map on top, at the root and inside each folder, points the AI to the right one. Nothing here runs. It's just knowledge sitting in its canonical place. The moment something becomes a real repeatable process, its process folder links to that knowledge. Never copies it in. So the tree ends up with two kinds of folders: containers that hold and route, processes that actually do things. Same rule either way. One home per piece of information. Still shaping this as I build it out, it's actually the backbone of what I'm working on: a base container structure any business could start from.

👤 Who I am: Izumo Spedicato, Italian, based in Malta. Explorer, Software Engineer, Digital Entrepreneur, Investor. 🛠️ What I actually do: I build for my company, Curiositas: https://www.curiositas.studio/en 🤖 What I'm building with AI right now: I'm building my company's first product: a portable, owned AI-operating container (law + memory + tasks + business departments) that gives an owner one folder that governs how they work with AI, instead of another SaaS to rent. The container itself follows a 4-layer ICM structure (no stage-contract layer, since it isn't a repeatable process), and it lets you spin up full 5-layer ICM workspaces inside it for anything that is. 🎯 What I'm looking for connection-wise: 🤝 A collaborator / accountability partner 👀 Just here to learn from people in my field 🧰 Trading workflows & systems 📬 Best way to reach me: DM here, or LinkedIn: https://www.linkedin.com/in/izumo-spedicato/