Vibe coding pulled me into a rabbit hole today. I am on day seven of a 30-day experiment. The aim is to take a business concept through build, launch and delivery, and hopefully to profit, using AI for 90% of the workload. Day seven was frustrating, and the lesson from it is worth writing down.

Here is what I did. I spent most of the day building the strategy and objectives side of the system. I already have task and project management in place, but projects need to come from somewhere. They come from your business purpose, your values, your objectives and your plan. So I built an AI wizard to hold your hand through that process. It asks a series of questions and you dictate your answers. It then fills in the whole document. The document runs from nine years out, to three years, to one year, and then breaks the year into quarters. The quarterly targets become projects. The monthly milestones then push into the task and project setup. The loop closes.

When I look at the output, the progress is big. When I look at how the day felt, it was a different story. There was a lot of testing, tweaking and iterating. Each change was small, and each one took time.

This is the worm hole I want to name. When you vibe code, changing things is so easy that you never stop. You notice a micro detail that most people would never see. You know it exists, so you fix it. Then you notice another. The ease of the fix hides the cost of the hour. I did exactly this today, and I failed the 80/20 rule. I should have asked which work gives the biggest return, done that first, and left the dotting of i's and crossing of t's until the end.

There was one clear win. Page load speed was slow, and it frustrated me. I worked through it with Claude Code, and the pages now load instantly. A fast page changes how the whole product feels. That fix earned its time.

I also have a weakness to admit. When I focus hard on my own build, I forget to check the work my team has finished. I need to bring their outputs into the overarching system, and I have not done it yet. One team member is building the sales and marketing side. Another is testing the user experience and hunting bugs. One has used Claude Code to build a new brand style. I am on product build. Because I created the systems and processes first, the three of us work in parallel without waiting on each other. The risk is on my side. Their work sits unchecked while I stay heads down. So tomorrow I review the website copy in the new brand style and read the testing report.

Where am I against the plan? Day seven of 30 puts me close to a quarter of the way through. I am ahead of schedule. Version one of the website is built. Two thirds of the operating systems exist in version one or higher. None is ready for deployment or sale. We are now moving into user interface and customer experience. Another system is close to version one, and I am joining them together so they work as a bundle. I have not packaged anything for sale yet, so I will not claim more than that.

Why does this belong in a running diary? Because the lesson is the same one the streak teaches me. Today is consecutive day 2,151 of Diary of a Runpreneur. My cumulative distance stands at 16,520.44km, with 23,554.56km left to reach 40,075km, a lap of the world. I run barefoot style every day, and I vlog every day. None of that works through intensity. Pace has to stay sustainable for the distance. A runner who sprints the first mile of an ultra pays for it later. A builder who polishes every small detail in week one pays for it in week four.

Consistency over intensity applies to the work as much as to the road. Daily effort on the right things beats a long day spent on the wrong ones. The 80/20 rule is pacing for a business. Put the effort into the work that moves the outcome. Save the polish for when the structure is sound.

Behind the streak is a mission. I run and vlog to raise £1M for children's causes. The more people the vlog reaches, the more we raise. A project this size needs the same long view. Some days produce a lot of visible progress. Others, like today, produce progress and frustration together. I keep going on both kinds of day.

My takeaway from day seven is simple: - Do not get stuck in the minor detail. - Do not lose sight of what you are trying to achieve. - Check your team's output before you build on top of it.

If you are building with AI, or thinking about it, follow along. I will share day eight tomorrow. If you believe in the mission, subscribe, share and comment. Tell me where you fall into the detail trap.

Watch the full episode: https://youtu.be/pe8sO1NDETY