Companies waste resources building mediocre in-house AI when they should integrate best-in-class third-party services instead. The real challenge is making products work seamlessly with AI systems through practical interoperability: copy-to-clipboard, markdown output, and document export let users move data between tools effortlessly.
Chapters
Enable third-party AI. You know, I find there are a lot of companies that are running around trying to add AI to their products. Most of the time, it's actually pretty crappy AI. They're trying to keep their costs down, trying to keep their tokens down, so how they're SaaS, and they're worried about having a meter attached to it, etc. But really what you want to do is attach the absolute best-in-class AI to the system. You want to help your clients get maximum value from your product or service offering by using it with the AI they would want to bring to bear, which also might be very different from the one that you're going to go build for yourself. So how can you compose with AI? I think this is sort of the new API problem is going to be how to be making these things work with AIs that like want to speak in the relatively, you know, natural language or unstructured data format. This is why I really favor, you know, enabling copy to clipboard. But even more than that, thinking in terms of markdown output, which is the natural language of AI, being able to export documents that sort of can easily make their way across. This is going to massively increase the value of your offering.