Hi guys, I’ve been working with my ICM folder structure for a while now, and I’m at the point where I want to make it accessible to two other team members as well. The main question for me is whether live synchronization and parallel work in the same folder structure actually makes sense, or whether it would be better to split the folder system by each person’s workflow or responsibilities.. 🤔 Has anyone already tried a setup like this for a team or found a way to structure it?

This is a recording from my VIP sessions in my community I usually don't post these online so that we can talk about personal business things, but I felt I wanted to share this one. I will never be sharing these publicly again. For my VIP members if you want your files I promised they are uploaded at the bottom of the drawing room post in a ZIP here: Session 8 6/13/2026 - The Drawing Room (VIP) · Clief Notes Time stamps: 00:00 Welcome: the Ledger and the ICM deployment layer 02:11 Member intros and what people are building 04:07 Engelbart, 1962, and software as collaboration 08:15 Fable pulled, and why output beats features 15:18 Getting unstuck on ICM 19:47 The three questions and a live ICM routing demo 33:08 AI as your runtime, humans in the compute layer 38:40 Productionize your opinion 41:40 Ingest agents and distilling your brain into files 46:39 You are not behind 49:23 London Tech Week: the rooms and the money 52:27 The buyer is changing: selling to agents 54:08 Everyone is overbuilding, and the talent layer opening 61:13 Placement fees, freelancing, and the college problem 66:47 No "best," and a bet on humans 70:05 Chicago: hollow output and the 60/30/10 rule 73:53 When to hire a human instead of automating 78:27 SkillOpt: training your skill files 90:29 Launching this week, and close

💡Let me start with the truth of things first, because the truth is the lesson. ================= Token Warning ⚠️========== Before anyone comes for me 😅 📝 Note: depending on what you are loading, you are accruing token cost. And token cost can balloon if you are not careful. If any of the files in the initializing phase or handoff phase are large this can become a token intensive practice so when curating this process, BE INTENTIONAL only load what is absolutely needed for the project to start up and for you to get to the fun part, building! I budget for my own needs, for me this initialization for me is between 5k and 10k, I have made the mistake of making 40k, don't do that! If you want to understand an estimate of what your character to token ratio is use the site below: Tokenizer - OpenAI API Anthropic tokenizer has not been released yet, but Its a few % higher than OpenAI in some cases. Plain English sentence (115 characters) - OpenAI GPT-4: 23 tokens - OpenAI GPT-4o: 23 tokens - Claude (content only): 23 tokens - Result: identical Code and markdown snippet (86 characters) - OpenAI GPT-4: 24 tokens - OpenAI GPT-4o: 26 tokens - Claude (content only): 27 tokens - Result: Claude a few percent higher Numbers and punctuation (55 characters) - OpenAI GPT-4: 25 tokens - OpenAI GPT-4o: 25 tokens - Claude (content only): 27 tokens - Result: Claude about 8 percent higher One honest footnote: Claude's API adds roughly 7 fixed tokens per message for formatting. On a single short line that looks like a big gap. On a real document it amortizes down to almost nothing. ✅The takeaway: for the same English text, the tokenizers land within a few percent of each other. The scary "Claude costs way more tokens" headlines do not hold up when you actually measure it. Verified beats confident, every time. You have been learned and you have been warned! ============ Back to it! 👇=================

Every time you drop a PDF or Word doc into a model, you pay tokens for formatting it never reasons over. Layout, styling, table scaffolding. None of it earns its place in the context window. So I built to-md. It turns .pdf, .docx, .pptx and .xlsx into Markdown, and writes two files: a faithful one, and a lean one stripped to the minimal text. The lean file is the one you paste in. One engine, three ways in: a command line tool, a Finder right-click action (Adds the path of the lean file to your clipboard too), and a Claude Code skill that fires the moment you hand it a document. The principle underneath it: the model should only read what it has to think about. The rest is waste you are paying for. Pre-process the document. Then prompt. //A<3

I wanted to share a win and a lesson I learned with you all, I just landed two new clients. A law firm 🧑‍⚖️ and a real estate office 🏘️. Here's the part I keep turning over in my head about what actually happened today. 1️⃣ The law firm wants in on structured workflows, the kind of thing a lot of us are building here. ✅ 2️⃣ The real estate office? They want tools and clean, no-hassle communication, but they don't want AI within a hundred feet of them. Same week, two clients, opposite ends of the spectrum. ✅ 💡This really made me think in the car ride home, because it forced me to lead with the problem, not the product. My job here wasn't to sell AI, it was to solve the problem in front of me, with whatever best fit the friction. One was agentic the other non-agentic. (With AI and Without AI) I have always loved learning, at first surface level, then I am always pulled into the rabbit hole! Being a generalist, or an autodidact, means I can meet a client wherever they are, instead of possibly dragging them somewhere they don't want to go. And having the domain knowledge, knowing THIER WORLD well enough to actually help, that's the thing that closes the deal. The presenting-it part? Still working on that one. 😅 🤔 One of the things I actually want to pass on isn't about either meeting. 💡 It's about the time spent before them. I've been researching how law firms and real estate offices operate, I study trusted industry peer verified sources, I sign up to industry newsletters, I go to networking events, and I get uncomfortable sometimes yes, still I go. In this case, since I can't bring you all along with me as I go, I will share some of the sites below: Note: Relevancy is a huge deal here, just like any other citation, if you are quoting a fact, gate, or trend in a vertical and it's from 2013 and its 2026, there is a good chance that in the 13 years that have passes, something changed. Law Office research - Make sure you are looking up the State Specific Sites in your area: