#TOKEN-MAXXED

We all know AI as a great assistant, one that can help you brainstorm and generate creative results from a wide array of data it's been trained on. But here's what we often miss: it's not always going to give you the best results. And that makes sense, actually, because AI doesn't have judgment. It can only provide possible solutions based on your input (context, data, prompt). The amazing results we get are usually slightly above average, but they appear great, especially if you know very little in the area you're working on. This in itself is not a bad thing. I'm not trying to point out a flaw in AI. I'm trying to highlight an important insight. YOUR EXPERTISE ALWAYS MAKES THE DIFFERENCE If you've ever worked with AI on a subject you're an expert in, you probably know what I'm talking about. When your expertise is actively influencing your work with AI, the results are directly proportional to your level of expertise. This is where you witness the full power of AI. When you help it navigate the right trained data based on your judgment and expertise, you create unique results others can't replicate. This is why I always say: focus primarily on your industry. There's a lot your years of experience can unlock, and it's a win-win, because those with no plans to switch careers get to deepen their work in the field they know best. I believe everyone should work with AI across three layers: 1. Primary AI work, where you use your expertise and experience to create unique work and improve your industry. 2. Secondary AI work, the auxiliary functions needed for your primary AI work, business, projects, and other important tasks. 3. Tertiary AI work, which powers your hobbies, entertainment, and other luxuries. This is how industries maintain their relevance, prosperity becomes attainable, and the quality of the human experience is preserved in these times. What kind of work are you using AI for?

This might not be sustainable since these days we want every thing to be a workflow we never have to return to. But I thought to share because someone might need this. These days I tend to write down thoughts as they come and this practice has helped me write more. It’s a refreshing experience and I’m grateful for that. I remember the early days of grammarly. I used grammarly to structure my sentences and correct any typos or grammatical errors. But now everything goes through Claude. So if you can follow through with writing your first draft, here’s the hack. AI is great at refining drafts to make it clearer and preserve your idea without drifting. But it might introduce words and phrases outside of your voice. It’s even worse if it’s the AI voice that we don’t like. 💥HACK: When refining your draft, tell it to retain your voice and choice of words as much as possible except in times where there’s an unclear message, grammar issue or wrong choice of words. And for every change made, it has to give a summary of what changed and why so that we can apply judgment. If we think about, this doesn’t have to be a one-off exercise. You can begin to train your Copy agent based on these types of sessions and in no time you have a system that writes in your voice.

When I joined this group on July 1st... I had no idea that it would change everything for me. I was close to giving up on AI and multi-agent cluttered messes... Repeating my self daily, copy and pasting back n' forth from agent to agent, and getting no where, fast. It took weeks for ICM to "click" for me and I put a video up about it and Jake shouted me out here. What I learned that work for me: Stop trying to jam ICM over my clutter, and instead just start from the ground up the right way. If you've been knee deep in AI and you're trying to jam the shape of ICM into what you've already built, I would suggest you just build your ICM the right way from the ground up. I did that with 1 folder and it runs several businesses, products, apps, sites, clients, productions, blogs, etc. If you're brand new, even better, build it right the first time. You only need 1 folder to get started. Once you have the foundation with your 1 agent, then you give that agent tools: i.e. ChatGPT might be your 1 agent and Claude Code is a tool you give it to build. (I don't even use chatgpt but recently helped a client set up this way) Then you can add on top of it, grow it, and keep it well organized. This would be 50'ish days later since finding ICM and about 30 something days since the update to my first video when it "clicked" and... it's a culmination of 90 ish days since I first paid claude code $20 on June 11th. Questions? Comments?👇

I am pulling records about the same people out of a few different systems into one folder, so an agent can actually work with them. The merging is not the hard part. The hard part is that the same person shows up three times, slightly differently, and every system is convinced it is the right one. Phone written four ways. Name spelled two ways. One system has the newest email, another has the newest address, and the one with the better data is not the same one twice in a row. Right now I pick a winner per field instead of per record, and I keep the losing values in the same file so nothing disappears quietly. It works. But the file keeps getting fatter, and I am not sure fat but traceable is the right trade for something an agent reads on every request. For anyone whose folder is fed by more than one system: do you settle the conflict once on the way in, or keep both versions and let whatever reads the file choose at the time? And what happens on the day one of those systems changes its mind about a person you had already settled?