Most experts put their best thinking somewhere their customers have to remember to open. Typically, that is a course, a book, a PDF, or a separate app. Which is great when they are learning, but sucks when they are in the line of fire. A hard email lands. A meeting goes sideways. They have to decide what to do next. With the traditional course or app, they have to stop the work, log in, find the right resource, and try to carry it back to the problem in front of them. Most people do not do that. They ask ChatGPT. Or Claude. Or they make the call with whatever seems right at the time. The hub for knowledge work For a lot of people, ChatGPT or Claude is becoming the hub for knowledge work. Not just where they ask a question. Where they direct the work. The calendar, email, Docs, project-management tools, files, and task lists are still there. But they are moving into the background. Instead of going tool to tool just to move an idea forward, you tell the chat what needs to happen. Then connected systems can pull in the context, do the work, and bring the result back. What this changes for experts You cannot just deliver static resources customers have to remember to check out. Your method needs to be available inside the tools where they already work. A customer brings a real situation to the chat. Your expert app can bring the questions that matter, the framework that applies, the constraints they need to respect, and the next piece of work they need to create. The customer still has to make the decision and do the work. But your method can show up while that work is happening. That is very different from putting a chat box on top of a course library. Start with one real moment Do not start by asking what kind of app you should build. Ask where your customer keeps getting stuck, making the same decision, or needing your guidance when you are not in the room. Pick one of those moments first. Copy this into ChatGPT or Claude: I have a framework that helps [who I help] achieve [the result].

Why people keep going back to AI People are increasingly going to AI instead of experts when they want to get something done. I get why. You ask it a question. It gives you an answer. You ask a follow-up. It keeps talking with you until you feel like you have a plan. That is a lot more appealing than buying a course, reading a book, or trying to track down the right expert. The problem with "good" advice But here is the problem. General AI does not know what matters most in your situation. It has safety rules. It follows instructions. It works from patterns it has seen across a massive amount of material. By default, it will usually give you something that is plausible, familiar, broadly acceptable, easy to explain, and close to the patterns it has seen most often. That can still be better than where you started. It can sound smart. You can get a useful idea. But it may not get you the result you came for. Because transformation was never only about knowledge. Most people do not need another list of things they could do. They need to know what matters first. What they can skip. What applies specifically to them. And what to do when the first thing does not work. General AI does not know your method. It does not know the mistakes you learned the hard way. Or which exception matters in this case. Then it gives you too much You ask for help and get ten ideas. Then ten more. Then a plan, a checklist, then a research doc 25 pages long. You feel like you are working because you have so much to read. But you have not moved. Books and courses cannot adjust Books and courses have a different limitation. They are broadcast. The author has to make assumptions when the book is written. The course creator has to make assumptions when the course is built. Then the customer has to figure out where they fit inside that fixed path. A book can be great. A course can be great. But neither one can stop and say, "That part is not your problem. Skip it. Do this instead." What an expert app changes

That may have been my biggest takeaway from day three of Funnel Hacking Live. Not every session. Not a full event summary. If you want all of it, buy the replays. But these are the three ideas I keep coming back to. 1. A digital employee needs structure. Most people are still using AI like a chat box. “Write this.” “Research that.” “Give me some ideas.” That can be useful. But it is not the same as giving AI a real job inside your business. Russell showed how their team evolved from specialized agents, to management layers, to EA: a larger agent with the right context across the company. The important part was not just access. It was permissions. Some people can ask questions and pull reports. Other trusted people can authorize work. Your AI needs context, responsibility, guardrails, and clear limits on what it is allowed to do. 2. A software membership is hard to sell when you lead with “membership.” This part was very validating for me because it is basically Linchpin. Do not try to sell a big, vague software suite from the front end. Lead with one valuable feature and the specific problem it solves. Russell used Dream 100 as the example. Someone may not wake up wanting another marketing membership. But they may want a better way to build their Dream 100. So the VSL leads with that outcome, while the full software suite is the offer. The same is true for a MIFGE: something useful but incomplete. And for dramatic demonstrations like webinars, challenges, and events that give people a reason to show up and see the value. Build the system. Lead with a specific result. Get people into continuity. Help them use it and get success. 3. You do not have to solve every limitation in yourself before you can move forward. Rich Shefren talked about his pattern of getting projects about 80% done, then getting pulled to the next shiny thing. So he built an AI agent intended to help take the project the rest of the way. We can keep doing the personal-development work. We should. But we do not need to let a known weakness become an excuse for letting important work sit unfinished today.

Most people use AI backwards. That was one of my biggest takeaways from day two of Funnel Hacking Live. This week, I’m sharing the three ideas from each day that I think are worth carrying forward. Not every session (too many to list!). Not a full event summary (catch the replays for that). These are the ideas I think could change how you build, market, and grow your business. Here are my three biggest takeaways from day two. 1. AI needs more than a one-line prompt. Russell showed his new process for creating VSLs. And it wasn't just that “AI can write a VSL.” We've been doing that for years already. The question is whether it writes something worth using. Russell described a 10 / 80 / 10 process. You put the first 10% in up front. You decide what the outcome needs to look like. You find examples. You explain what you like about them. You give it your voice, your real stories. And swipes of the best VSLs in existence. Then AI can do the 80% of the work. At the end, you come back for the final 10%. You read it. You fix what is wrong. You add the judgment only you can add. He also gave a ratio that is worth thinking about: 500 words of context for every one word you want back out. That sounds extreme. But it explains why most people get generic AI output. They give it almost nothing. Then they expect it to know their market, their voice, their stories, their offer, and what a good result looks like. It cannot know what you do not give it. 2. Your social media is already a funnel. Brian Mark gave one of the most useful tactical talks of the day. He said his content works in three layers: Connection content helps people know you. Value content helps people like you because it solves a problem. Social proof helps people trust you. But the part I liked most was his self-audit. Pull up your last 14 posts. Then ask yourself: If my ideal customer found my page today, could they get a meaningful result without hiring me? If the answer is no, that may be why the content is not converting.

What if the way you work is already outdated? Russell opened Funnel Hacking Live with a simple question: What if? I have been thinking about that question ever since. This week, I’m going to share my top three takeaways from each day. Not every session (I missed some). Not a full event summary (buy the replays). The things I think are worth carrying forward. Here are my three biggest takeaways from day one. 1. Start with “What if?” Russell opened with that question. What if the way we think about work is already outdated? We are used to output being tied to human hours. You hire people. They work. They produce. That is still true. Someone still has to decide what matters. Someone still has to give direction. Someone still has to check whether the work is right. But AI is changing how much one person can get done. What if you could build software without needing to hire a $150,000 developer before you can test the idea? What if you could build something useful, put it in front of customers, and learn from it before making a huge bet? That is a very different way to think about starting and growing a business. 2. ClickFunnels appears to be asking a simpler question again. ClickFunnels Classic was focused on funnels. Then ClickFunnels 2.0 added a lot more. More features. More tools. More things to do. But one of the questions that came up at FHL was: What if we stripped it back? What if we got really focused on funnels again? There are still a lot of unanswered questions about what ClickFunnels AI will become, how it will work, and what it will cost. I do not have those answers yet. But I came away thinking the conversation is moving away from: “How do we add more features?” And toward: “How do we help people build the thing they actually need?” 3. Courses are not dead. But static delivery has limits. Richmond Dinh gave one of my favorite presentations of the day. His point was that people are going to AI because it can feel more certain than an offer that says: