Today's Innovation is Tomorrow's Table Stakes

I wrote one of the first content management SaaS apps, back before the original dotcom bubble burst, before Wordpress... it was new and innovative and while we were not the only one, there were only about 5 others. We did a decent job, for 1999. 

Fast forward 10 years, and writing a CMS is a weekend project while learning to code. Maybe not a good enough one to turn into a viable product, but conceptually it is a basic learning exercise. And the market is flooded with them in all shapes, forms, and levels of quality. 

Dashboards and analytics were next.  Really fresh and new at one point, with a few vendors offering them. Then the market flooded because everyone can do it. And now if you don't add some dashboard features in your app, you are not a viable product.

There are other examples - those are just the two I felt directly. But the bigger point is that what is new today is only new and innovative for a short time. Once it has been done, the concepts are understood by the rest of the industry and they become commonplace. 

We are at the start of that process with AI. A year ago everyone was trying to figure out what to do with the new capabilties LLMs bring to the industry. And now, it seems that all products have it. When it is as easy as writing an API call and sending some text, why would you not integrate AI into your products? It is both a highest impact change in our industry and easy to adopt. 

Because of that, just adding AI is not innovative - It is table stakes - if you don't have it, your product is just a prototype, not something that will find a market.

The product managers who will come out on top of this change are the ones who accept this and correctly apply AI to solve the problems well-suited to it. They will build a deep understanding of what AI is good at, and what it is not. The current wave of feature development is trying to make AI do all things for all people, and it will not last. A year from now, I predict we'll have solid documented answers of the strengths and weaknesses of LLMs, with known subtleties and nuance for the various avilable models.

So to be at the peak of innovation in 2024, you need to be ahead of that learning curve. You need to know when to say: "Yes, implement that AI feature" vs. "No, don't do it - it will just bring poor AI results into our product." You'll need to keep up on what new models are coming out and how they change those answers.  You'll need to understand which AI outcomes delight the users, and which make them turn away because it is just AI-created nonsense. In short, you'll need to be laser-focused on delivering features that solves problems well, and not banking on people being excited just because AI is shiny.

If you can do that, and build products that quickly deliver the latest AI strengths to solve specific problems, you'll be at the edge of innovation.

For a year. 

Next year, there will be something new.