Microsoft Word was once the most commonly used software in the world. A .doc file was the lingua franca of the computing world, and “send me a Word doc” became part of the business lexicon. Word won the battle against WordPerfect, which was never quite able to make the transition to the world of Windows. 

That battle with WordPerfect might have been a pyrrhic victory, however, as Word ended up something quite different than what the original product manager might have hoped. By out-featuring WordPerfect, MS Word became a bloated and unwieldy application that had way too much stuff jam-packed into it. It fell victim to the “just because you can do it doesn’t mean you should” syndrome. Each new release included more obscure and less-used new features that looked good on a marketing sheet, but that only made the product more confusing to end users.

And all that happened in a world where new features had to be coded by hand and took weeks or months. What is going to happen to software now that adding features can be done with AI in an afternoon?

Highway to featuritis

Software product managers have a challenging job. One of the biggest difficulties is central to their role: What features get added next? Adding features usually takes time, and thus a backlog of features accumulates. This gives the product manager time to vet features, examine them, determine their fit for the product, and ultimately decide if the feature is worth the effort. Items in the backlog are constantly evaluated and reevaluated to determine if they will make the product more appealing to customers.

In other words, the existence of the backlog gives the product manager time for proper due diligence. But with the advent of agentic AI, the days of features languishing in the backlog are coming to an end. Agentic coding will allow features to be conceived in the morning and shipped in the afternoon. Our build and test pipelines already allow bugs to be fixed and deployed in hours. We are about to experience the same acceleration for product features.

And this presents a new challenge to product managers. Instead of having to decide what well-vetted features to build next, they are going to have to make rapid decisions about whether a given feature is worth doing. 

The temptation, of course, is to add as many features as possible, because the competition is certainly already adding them as fast as possible. And this puts us back into the situation where “featuritis” or feature creep threatens to bloat and overcomplicate a product — something that good product managers are careful to avoid.

Coding unleashed

The problem is made worse by the fact that developers can add features so quickly that they can — and probably will — bypass normal processes and just add the feature without anyone stopping to ask if the feature is valuable, desirable, or even useful. Those processes — which take into account security issues, legal factors, and market forces — exist for a reason. Bypassing them can have serious ramifications. The challenge shifts from not having enough time to build what you want to not having the time to decide what not to build.

This will require a cultural shift in organizations. Product managers will have to shift from trying to convince their organization to squeeze one more feature into a product cycle to trying to keep superfluous features out. Instead of being pressured by upper management to add more features, forces will start to muster to limit the ability of teams to add features just to keep things under control. 

It used to be the hard part was jamming in that extra feature. Now? The hard part will be keeping them out.