Financial clarity is the thing most business owners say they want and few have. Today I reached day 14 of my 30-day AI experiment, and the accounts payable work I finished is why I am writing about it.

The experiment is simple to state. I am taking one business concept to profitability within 30 days, with AI doing around 90% of the workload. That is the target, not a result. Halfway through, I am still on track, and I have no illusion about how much work remains.

Today was also consecutive day 2160 of my daily barefoot-style run and vlog. I have now covered 16,584.61km of the 40,075km lap of the world. That leaves 23,490.39km. I run and film every day to raise £1M for children's causes, and the more people the vlog reaches, the more I can raise. That mission sits behind everything else I do, including this build.

What I built today

I built two accounts payable modules, one for fixed costs and one for variable costs. They connect to the rest of the system, and AI does much of the grunt work of interpreting what comes in. The aim is real-time financial data that updates as the business runs.

I also built out the credit control process. A large share of it is now run by AI. If money does not arrive when it should, the right steps start and the right workflows run to recover what is owed. The owner no longer has to remember to chase.

I finished the external contractor piece as well. People working in the field rarely want to log into a complex system on a phone. So I built a Slack interface where a contractor talks to the software through a simple channel. It is a small change, but it matches how those people already work.

Why I built finance this way

When I mention this, people say they can do the same in Xero. They probably can. In my experience, though, most owners leave Xero to their accountant. The terminology suits an accountant and does not suit the client. I wanted something an owner reads in seconds and acts on.

Most business owners and entrepreneurs do not have clear sight of their own finances. That is a risk to the business. If you cannot see what you owe, what is owed to you and what is coming, you decide on a guess. This build puts that information in front of the owner, with AI adding insight and suggestions on top.

There is setup behind the scenes, mainly the database. Once that is in place, templates make repeat setups far quicker. I am not claiming it is effortless.

The pressure that comes with progress

I started this as a relaxed experiment. I wanted to see what would happen. Two weeks in, the product is more complete and the idea looks more like a viable business. That brings pressure I did not feel on day one.

With software, reliability is the real test. I fix bugs and gaps every day, and I want to do as much as I reasonably can before the first external test clients arrive. My plan for the next week is to keep averaging around three modules a day. The final five days are for migrating everything to a more robust infrastructure that I can hand to beta testers.

I also have other businesses to run, so I cannot give this every hour. In practice that works in my favour. I use the Claude Max 5x plan, which gives me five-hour usage windows. I can get through a window in two to four hours if I run several sessions, which leaves pockets of waiting. I could move to the 20x plan and I probably will later. For now I want to show that you do not need an expensive subscription to run an experiment like this.

The lesson for me

The more time I spend on something, the more I understand it. The more I fix, the better it gets. The better it gets, the more I care about it, and the more it looks like a real asset. Time and the first test clients will tell me whether I am right.

This is the same pattern I see in the running streak. Day 2160 is not the result of one big effort. It is the result of showing up on days when I felt relaxed and on days when I felt pressure. Consistency matters more to me than intensity, and I want to keep a streak going for 16.5 years in total.

Where this leaves me

For any business owner curious about AI, my advice is to start with one real problem. Mine was financial clarity and credit control. Build that part properly, test it, and fix what breaks.

Tomorrow I run again, build again and film again. If you have a question about the build, the finance modules or the running, leave a comment and I will reply to anything relevant. If you believe in the mission to save children's lives through this ultra-marathon of consecutive daily vlogs, subscribing, sharing and commenting helps the vlog reach more people.

Watch the full episode: https://youtu.be/L-OvvCvAU58