I found myself completely stuck today, dreading to continue working on the business plan from the last couple of days. I don’t want to work on it today. Yesterday I had the insight that my business plan is based a lot by me having to knock on doors to sell local AI services. The thing I just don’t want to do is having to sell face to face. I prefer just having a fully online business instead.

Of course this realization could be countered by just busting through that resistance and not pussy out. However there is so much opportunity out there, that I want to do something I actually get excited about. My requirement checklist for the new business idea were as follows:

  • B2C -> I want to have lots of customers, to reduce key account risk.
  • A product -> Fun to have something in your hand that you made.
  • Start in the Netherlands, but possible to scale worldwide
  • Have a low price point
  • UGC marketing possibility -> I see a lot of examples of this and want to try it out.
  • Don’t do fulfillment myself -> I want something without my involvement when a sale is made.

So I started brainstorming new idea’s based on my previous experiments and these requirements.

I have 2 children and we love reading to them. Especially if the story resonated around a theme we want them to learn. So I came to the conclusion that building a custom story book business with AI generated characters based on real photo’s would be a good fit.

There are some competitors, but the options available in The Netherlands are not that good. When I saw the resulting book it featured an image and a normal white page with the text. This could be done way better and with a more premium feeling.

So I started Claude Cowork to build a business plan around this idea. Usage limits again, so I switched to Codex. Let’s see if it is as good as the Claude business plans.

First I imported my llm council into Codex. As I already had a Claude skill file for this, but usage limits. After running Codex in /plan mode mentioning it should use the llm council I saw Codex spawn 5 subagents. That was new to me, and this gave me a feeling it would improve the results. When they were done counceling I had a fully fledged business plan made. And the nice thing is that it was very thoroughly examined by the council of llm’s. It showed in the plan, as it had a lot of nuances I missed earlier. For example the idea that this is a premium gifting company for adults buying for children and to start with one flagship hardcover product.

Key Insight:

When building something becomes very cheap with AI, excitement is the only real constraint. So having something that excites me is good, it will keep me engaged!