Fundamentals Still Apply

Every week, someone tells me their company needs an “AI strategy.” Yet they don’t - they need to remember how to build great products, and might have forgotten that a while ago.

From a product development view, AI didn’t create a new category of problems - it just magnifies the ones you already had. If your roadmap was a list of customer requests before AI, it’s still a list of customer requests - just now with “add AI” scribbled in the margins.

If you didn’t know what to build before, a new model from OpenAI won’t help you figure it out.

The process hasn’t changed:

• Who are you building for?

• What are they trying to achieve?

• How are today’s options broken?

This is the timeless formula for building great products. What’s new is the pace, the tools, the interfaces, and the fresh possibilities now within reach. Nothing more.

Here’s what that means in practice.

For your product: bolt-on AI is the most expensive way to become irrelevant. Slapping a chatbot onto your existing product inherits every constraint of the system it sits on - the data model, the UX, the architecture. The real question isn’t “where do we add AI?” It’s “what if intelligence was built in from day one?” That’s the scarier question, but the one that truly moves the needle - as it forces everyone to think radically different.

If you don’t ask this (and answer), somebody else will, and is very likely already doing so.

For your roadmap: stop asking customers what to build. Seriously. Customers see problems through the lens of your existing product. Their feature requests are symptoms, not strategy. What you need is instrumentation, conviction, and taste - not another feedback survey. The real answers lie in data, their behavior, and your observations, not in the interviews.

That one is - you guessed it - the same old story since the beginning of time.

For your team: you don’t need more people who write code. You need people who know what to build.
I’m not even diving into debates about “building is cheap,” “software is solved,” or “AI codes better.” That discussion is irrelevant at this point. Whether you ship faster, slower, or sloppier doesn’t matter if you’re building something no one wants.

For your pricing: seat-based pricing is dying, yes, we’ve heard it. But the proposed fix - usage-based or token-based pricing - is just backwards. Sure, for the SaaS provider’s outdated business model, that’s what’s necessary.

Notice something? That’s not the customers’ problem at all. This mindset shifts the burden onto them. Now they must track token usage, justify costs to their CFO, and budget for unpredictable monthly bills - things they likely barely grasp. That’s madness. Any buyer would gladly pay double for a fixed-price plan that eliminates this hassle. Price based on the value you provide (which always boils down to making things easier), not the compute you use. This isn’t new wisdom; it’s pricing 101.

(On a practical note: Yes, you need to build in a mechanism to make sure that YOU can protect your downside and not overspend).

For your strategy: disrupt your own product before someone else does. I see companies stuck in “AI pilots” for months - pilots that succeed just enough to feel safe but avoid real commitment. Meanwhile, two founders armed with an agent and domain insight built in six weeks what your team predicted would take eighteen months. The real threat isn’t your usual rivals. It’s rule-breakers, general-purpose agents like Claude CoWork and OpenClaw, and sometimes your own customers crafting their own apps. Stop piloting. Ship or be shipped - the good news is: Speed is available for you, too.

None of this is new. That’s the point. The companies that’ll lose aren’t the ones without some AI feature crammed into their product — they’re the ones that forgot how to build great products in the first place.