Google’s version if ICM? The **Open Knowledge Format (OKF)** is an open, vendor-neutral specification designed to standardize how context, metadata, and curated knowledge are stored and shared between humans and AI agents. Introduced by the Google Cloud Data Cloud team in June 2026, OKF formalizes the emerging **"LLM-wiki" pattern** (popularized by AI researcher Andrej Karpathy) into a portable, interoperable format. Instead of forcing AI agents to repeatedly search massive, disconnected text files or query proprietary vendor catalogs, OKF provides a shared standard for a living knowledge base. ### Core Technical Structure The technical shape of an OKF "bundle" is intentionally minimal and lightweight. If you have ever used tools like Obsidian, Notion, or static site generators like Hugo, the structure will look highly familiar: * **File System as Identity:** An OKF bundle is simply a directory tree of standard Markdown (.md) files. Each file represents a single unit of knowledge—called a **Concept**—which can be a physical asset (like a database table or API endpoint) or an abstract idea (like a business metric or an incident playbook). * **YAML Frontmatter:** Every Markdown file begins with a snippet of YAML configuration metadata. * **Minimal Schema Constraints:** To maintain ultimate flexibility, the specification enforces only **one** required field in the frontmatter: type. Other highly recommended (but optional) fields include title, description, resource (a unique URI pointing to the underlying asset), tags, and timestamp (ISO 8601). * **Graph via Hyperlinks:** Concepts are linked together using standard Markdown links ([Link Text](path/to/file.md)). This effectively turns a flat folder structure into a rich graph of untyped, directed relationships (e.g., *parent/child*, *depends-on*, *joins-with*). Broken links are explicitly permitted to represent knowledge gaps that haven't been written yet. ### The 3 Core Design Principles

Vote below, then tell us in the comments: have you ever created something at work or school that you do not own?

This is a little bit of a rant cause I can’t post these thoughts anywhere else. At least, I’m not ready to get publicly crucified yet, so… I cannot talk to almost anyone I know about ai. They have no idea what’s going on, or about the kind of stuff we do. It’s all doom and gloom. There is something that really bugs me about it, beyond the naivety and half-glass-full mentality. It’s that... gatekeeping does nothing for humanity and is a selfish mentality at its core. So ai has access to lots of data, and artists data. So what? If you’re really good, it doesn’t matter. Great products, great services, great talent all have one thing in common which makes them in-demand. They’re great. And greatness is witnessed by all the people who consume their work. There’s no gatekeeping because it’s literally on display. From Harry Potter to Michael Jordon, Disney to Taylor Swift, Steve Jobs to Crayola Crayons. If no one is trying to copy you, then your work isn’t loud enough. Either because it’s not good enough (yet), or because you’re hiding it away from the world out of fear. In any case, their success came from being great, NOT from gatekeeping information or their talents and trying to sell them to the highest bidder. Bringing it full circle… Nobody lost money cause I finally had a tool that could do work for me for free. Rather, everyone in my companies will literally make more money because of massively efficient operations. It also means we’ll be able to afford remarkable talent with the extra profits we bring in due to that efficiency. I’m SO glad I have more options to take care of the people I bring into my sphere. One last thought… a little side quest here.. Whenever I ask people about why it isn’t a good thing that robots will be able to farm all the organic food we need at extremely low costs to produce compared to what we have now, no one has a good answer. I used to be of the mindset that really rich people should use their money to fix problems like world hunger. Then I actually looked at the challenge of getting the recipients to use money the way it’s intended, and started crunching numbers to see how long they could solve it for. Full stop, it doesn’t work. Because complex problems require complex solutions. A billion dollars solves world hunger for how long… a day? A week? A month? And then what?

Everybody’s posting about what to do with ICM. Let’s take a moment to focus on things that may trip you up. There’s a lot of great content in here right now about what you should be doing with ICM. The builds, the workflows, the wins. I love it. But I want to talk about the other side of it for a minute. The stuff that quietly wrecks your workflow. The small things you don’t find out about until you’ve already learned them the hard way. Most of these I learned by doing them wrong first, so take that for whatever it’s worth. 1. Copying builds instead of learning from them: The builds people share in here are incredible. But almost all of them are built for a very specific situation. When you download one and start pulling pieces out to make it fit your world, you leave things behind. Orphaned rules. Edge case handling that made perfect sense in their context and now quietly fights your normal workflow in yours. You don’t see it because these things get intricate. It runs fine right up until it doesn’t, and then you’re hunting a problem you built into it on day one. What I do now is I don’t adopt a build. I study it. I want to understand what it’s doing and why it’s doing it. Then that reasoning becomes part of my toolset, and I build my own thing with it. You’re trying to import the thinking, not the template. 2. Skipping the part where you map your own logic: Before I touch a single folder, I sit down and think about what I would actually do if I were doing the task by hand. If this happens, I do this. If that happens instead, I handle it that way. And if it shows up some third way I didn’t expect, then I need to do this other thing. You write all of that down. That is the work. ICM is about being intentional with your context. You cannot be intentional about something you never thought through. Skip this step and you’re just decorating a guess . 3. Going too fast: This is the fastest way I know to make ICM fall apart. Everything is now now now, go go go, get it done. ICM does not reward that. The whole method leans on you slowing down enough to be deliberate. When you rush, you skip the thinking, you grab somebody’s build, you bolt it together, and you ship something that was never solid underneath. You have to be willing to sit with it.

At a going away party this weekend, I was talking to some government workers who mentioned they're mandated to use Copilot at work every day. Naturally curious I asked, "Oh what do you use it for?" They both sheepishly admitted that they don't. "We just don't see how it's useful for what we do." Later that night: birthday dinner at a German restaurant I'd been trying to get to for years. Some old theater colleagues: a props artisan, a lighting designer, they spend their days with hands-on physical work. AI didn't come up once. Meanwhile I was sitting at the table quietly asking Claude to translate the menu. It didn't just give me definitions — it told me the story behind one of the dishes. Monks during Lent, hiding meat inside pasta to get around fasting restrictions. Made for a very fun conversation topic but nobody noticed I had found out from Claude. It's very interesting being inside this bubble of thinking about how to make the most of AI all the time and finding that everybody else seems to just not care. A small win though, my brother got a Claude subscription this weekend. I've been sharing with him what I've been able to do with what I've learned here and it took him a bit to get over the hurdle of paying for a subscription (he's very frugal), but after a day auto-transcribing a jazz improvisation recording straight to sheet music, he's already organizing his context with folders and texting me about projects he wants to build. How is everyone else finding how people outside the community talk about AI?