Why we made this

These recipes try to capture some of the diversity we've encountered. We've deliberately included traditional machine learning techniques, things that only a few years ago would have been considered cutting-edge AI, alongside newer approaches like using Claude Code.

We'll update this as we learn more.

The world of AI moves quickly. It's difficult for any website to keep pace. Some techniques here, like the machine learning approaches, are basically evergreen at this point. Others are areas where best practices are still emerging.

Worth being aware of that. Tools like Research mode or web search within large language models can help you check whether anything has moved on.

We try to keep on top of this ourselves. Under each recipe you'll see the most recent review date, previous review dates, and who did the review. We do both human and automated reviewing.

The records of our automated reviews are at /debug/reviews. We've included that for transparency, but also because we think it's an interesting glimpse into how we're using AI as an agency to improve the quality of work we can do.

When we first wrote these recipes, we introduced a number of suboptimal suggestions. Gemini corrected us on several of them. We think this kind of human-machine interaction is worth showing.

We hope you find these recipes useful. We enjoyed creating and testing them.

Drop us a line at hello@make-sense-of-it.com if there are recipes you think are missing, or anything in the existing ones that could be improved.