One of the easiest mistakes teams make with AI is assuming more usage automatically means more value. Once people start seeing time savings, the temptation is to apply AI everywhere, to every task, every workflow, every stage of work. But the fastest teams usually do something more disciplined than that. They do not use AI for everything. They use it at the moments where work tends to slow down, stall, or loop back. That distinction matters because not every task creates the same kind of drag. Some tasks move fine without intervention. Others create delays, rework, handoff confusion, or blank-page friction that quietly stretches cycle time. The teams getting the best results are usually the ones that know where those slow points are and apply AI there first. ------------- More AI usage is not the same as better AI usage ------------- It is easy to think adoption success should be measured by how often AI appears in the workflow. But high usage on its own can be misleading. A team can use AI constantly and still save very little meaningful time if it is being applied in the wrong places. This happens when people focus on novelty instead of friction. They try AI on random tasks, experiment broadly, and generate a lot of activity without identifying where the real delays are. The tool becomes present, but not necessarily useful in a way that changes the pace of work. The better question is not, “Where can we use AI?” It is, “Where does work keep slowing down?” That is where time savings tend to become visible and repeatable. Maybe it is the first draft that always takes too long to start. Maybe it is the handoff where details get lost. Maybe it is the review stage where messy inputs create extra rounds of correction. These are not glamorous problems, but they are often expensive ones. Fast teams understand that the point is not broad insertion. The point is targeted friction removal. ------------- The biggest gains usually live at the stall points -------------

In this video, I show off Claude Dispatch, which is essentially a clone of OpenClaw but safer and easier to use. Anthropic wasn't alone though, as Perplexity and Manus both released OpenClaw copies (OpenClawpies?) themselves. I also break down this trend in AI along with the new Gemini-based updates to Google Workspace and Google Maps, and more. Enjoy!

Your life usually doesn’t fall apart all at once. It drifts. A little less focus. A little more distraction. A little more scrolling. A little less doing the things you know you should be doing. And over time, that adds up. I’ve learned this the hard way more than once. If you want to build something meaningful, you have to protect your focus like it’s your job. Because in a lot of ways… it is. Not every opportunity deserves your time. Not every opinion deserves your attention. Not every thought deserves to be followed. Stay locked in on what actually matters. That alone will put you ahead of most people. So, what are you focused on right now and what are you going to do this week to protect that focus at all cost?

In this video, I did something a little special, as I was out of commission for a week due to surgery. Instead of skipping the week in AI news, we put some of the best modern AI tools to the test to see what we could create. So I'm proud to present our guest host AI Igor, who will only be filling in this week while I rest my voice. AI Igor covers the results of the testing we've been doing on the top models for the past week, talks about the new Copilot Cowork coming to Microsoft 365 users, discusses the disappointing release from Luma with Uni-1, and more. Enjoy this special edition and I will be back next week!

A lot of people say they want teams to adopt AI faster, but many of the social signals around AI make adoption harder. The tool gets framed like a test of who is innovative, who is behind, who “gets it,” and who does not. Once that happens, people stop approaching AI as a workflow tool and start experiencing it as a referendum on their ability. That shift creates delay. It adds pressure where curiosity should be. It turns simple experimentation into a performance moment. And it makes the learning curve feel more personal than practical. If we want AI adoption to move faster and create real time savings, we need to stop treating it like a talent test and start treating it like what it actually is, a way to reduce friction in the work. ------------- Performance pressure slows practical learning ------------- When a new tool enters the workplace, people do not respond only to the tool itself. They also respond to the culture around it. If the unspoken message is that capable people should already know how to use AI well, then anyone who feels uncertain is likely to hide that uncertainty instead of working through it. That is where time starts to get lost. Instead of asking basic questions, people stay quiet. Instead of testing a small use case, they wait until they feel more confident. Instead of learning in public through normal trial and error, they try to avoid looking inexperienced. This is a common pattern in high-performing environments. People are comfortable being competent, not visibly early. So when AI becomes tied to status, speed of adoption often slows down. The people who most want to avoid wasting time end up spending even more time observing, second-guessing, and delaying the first useful experiments. The irony is that AI does not usually become valuable through image management. It becomes valuable through repeated practical use. And practical use gets harder whenever people feel like they are being evaluated instead of learning. ------------- AI is not proving who is smart, it is revealing where work is inefficient -------------