Today I wired the chatbot to a website to make a demo. Codex and I made the website that looked mediocre, but did its job. At the end of it I sat down to write this piece and didn’t know what interesting angle to write about. Well, this might be the thing to actually write about.

The difference between days that teach me something and days that move me forward. They both count, but feel totally different from each other. Today I am moving forward on the goals from the business plan in a breeze. They are well defined and easily handled by Codex without any hiccups.

  • Create a free Carrd website (one-page restaurant website)
  • Embed the Voiceflow chatbot widget
  • Test the full experience from a customer’s perspective on mobile and desktop
  • Screenshot everything — this becomes your demo material

In the business plan a Carrd website was proposed. However Codex found this to be incorrect as it could not integrate the Voiceflow chatbot in the free version. So we switched to a Netlify static website for it.

Then the some more technical friction came from Voiceflow itself. The public web widget kept failing with a Project not found response even though the underlying runtime was live.

It got done without too much fuss by Codex. This keeps happening, I have a problem. I explain it and from there it is smooth sailing without me doing anything, except for allowing all sorts of terminal commands to be run.

It started to get harder to actually write something today. As writing the same thing (started Codex, prompted my requirements, result got built, result was good) over and over again gets repetitive. So my write ups are becoming harder to keep interesting for my readers, but also for myself.

I seem to lose a bit of interest each time I build something and it just works. It makes me rethink my commitment of doing an AI experiment for 365 days.

I went through an exciting phase when I started to do the daily AI experiments. I had a holy shit moment when I experienced that AI can build things I have never been able to build myself.

I then hit the peak as these experiments started to write themselves. And now I feel I am stuck in the middle, where it just works. And “it just works” is actually the hardest thing to write about.

Key Insight:

The AI results became ordinary. And ordinary is where real businesses get built. It is also a terrible place to write from when chasing novelty. I need to get out and talk to real people in restaurants and pitch my demo!