Launch a business with AI. That is the question I put to myself this week, and I am going to answer it in public, one day at a time, over 30 days.

Today is day 2141 of my daily run streak. I have now covered 16,448.18km of the 40,075km lap of the world I am working towards, which leaves 23,626.82km. I run in barefoot-style footwear, I vlog every day, and the aim is to raise £1M for children's causes. I mention this first because the way I run is the way I am approaching this test. Small, daily, recorded and honest.

Why I am running this test

I spent 18 months building a software consultancy, ready to launch. At the last minute I pulled the plug. AI was changing the space so quickly that work which took me 18 months could probably be rebuilt in a week or so. The value of what I had built had dropped, and carrying on would have been a decision driven by effort already spent, not by evidence.

So I changed direction. I already had years of experience in systems, processes, dashboards and business infrastructure. I decided to turn what I had built into operating systems for small and micro business owners, sold for a one-off fee. The buyer gets a web app they own for life. I host it and look after it.

My reasoning is simple. Costs are rising and many small businesses are tired of adding another monthly subscription. A single fee is easier to say yes to.

Where I stand

I want to be accurate about my starting point. I am not at the forefront of AI. Plenty of people are doing this already, and I am not late, but I am not a pioneer. I would say I am above average in how I use AI, and a little ahead of the median. I know systems well. That is the advantage I bring, and it is a modest one.

Yesterday I suggested I might launch a business in a week. Having thought about it, a week is optimistic. Thirty days is a fairer test, so that is what I am committing to.

What I built in one day

The first operating system takes an idea through to a business plan and a 30-day launch plan. The user answers questions. I have no involvement at all. The system stress-tests the idea, builds financial projections, carries out market research and produces either an HTML dashboard for a web browser or a PDF export.

This work normally takes a lot of time. A half-hour question and answer session now produces something a person can work from alone or hand to a team. I built the first version in around three or four hours.

I will be straightforward about the cost. I am now so absorbed in building that I am giving it spare time on top of my working hours. So yes, you can probably do this in 30 days, but it takes real commitment. Anyone who tells you it takes no effort is skipping the part that matters.

The rules of the test

The team is me plus two others. The brief is that AI builds, sets up, launches, operates and delivers the business. The humans oversee, point it in the right direction and grant access where AI needs it. My minimum target is that AI does at least 90% of the work.

We have identified 36 operating systems to build. Many will wait. We aim to launch with two or three and add more once the website and commerce platform are in place.

The costs, with the caveat

Fixed costs should come to around £6 a month. That figure needs context. I already pay for subscriptions in my existing business and I am using them for this build. The team is also paid through that other business, so I am borrowing resource. I do not think this removes the value of the proof of concept, but you should know it. At those costs, one sale makes the business profitable. The objective is a profit after 30 days, and I am putting no pressure on it beyond that.

The lesson for me

The 30-day launch plan is built day by day and spreads the workload across the three of us. Looking at it, I noticed how closely it matches my running. I have not run 2141 days in a row through intensity. I have done it through a plan that asks for something small and specific each day.

A launch works the same way. Day one is not the launch. Day 30 is the sum of 30 ordinary days of showing up. Consistency over intensity applies to both.

The second lesson is about perspective. It is easy to be impressed by how fast a tool builds something. A fast build is not a business. A business needs a customer, a sale and delivery. I have a plan and a first system. I do not yet have a customer. I will keep that distinction in front of me so I do not mistake speed for progress.

The third lesson is about protecting what matters. Getting this absorbed in building is a risk to sleep, family time and the streak. The run comes first every day, because the run is what the whole mission rests on.

Why this connects to the mission

This vlog is part of my attempt to run the ultimate ultra-marathon, a lap of the world at 40,075km, in daily steps over about 16.5 years. The more people who see it, the more I can raise for children's causes, with £1M as the goal. The business test is separate from the fundraising, but both are built the same way, one day at a time and reported honestly.

What happens next

Tomorrow I present the business plan and 30-day launch plan to the team so each person understands their role. I will report what works and what does not.

If you are thinking about starting a business or using AI to improve one you already run, follow along. If you have a question, leave a comment and I will answer where I can. If you support the mission, subscribe, share or comment on the episode. It helps put the run in front of more people.

Watch the full episode: https://youtu.be/uCn303jOwNQ