Just a little reminder, since the world seems to have forgotten:
You can still build innovative, valuable technology without stuffing AI into the end product.
Really. It’s still allowed.
In fact, we’re in a golden window in which AI can turbocharge early innovation.
Use it as a research assistant to surface tools and techniques you didn’t even know existed.
Use it to brainstorm creative solutions to real problems.
Use it to learn new libraries and approaches, or simply to talk through the logic of what you’re trying to build.
You can even use it to code a prototype. I wouldn’t recommend shipping production code that way, but if you’re trying to get the first version of an idea off the ground, that’s where AI can shine.
Start with the value
AI is expensive.
But if you’re using it to reduce costs—by reducing expensive, time-consuming tasks, improving system resilience, or cutting down on tedious and error-prone research work—that is where it can really earn its keep.
This is especially true when you’re dealing with processes that are labor-intensive, require deep specialization, or are simply slow to do well and consistently. If AI can make those cheaper, faster, and more reliable, that’s a powerful lever.
You can save real time and real money while freeing people to focus on work that actually moves the needle.
Where you run into trouble is when you use AI merely to make something seem innovative without solving a costly or painful problem.
AI-related regulation is already expanding across the states, and it does not apply only to chatbots and language models. Depending on the jurisdiction, it may cover products that generate pricing recommendations, filter résumés, or quietly pull other consequential levers behind the curtain.
So unless the product genuinely needs AI to deliver value or reduce cost, you may be dragging yourself into a regulatory swamp for no good reason.
I love AI.
I’m also exhausted by AI.
It’s brilliant—sometimes even downright magical.
But lately, it feels like folks are trying so hard to make things “smart” that they forget to make them valuable.
Using AI to help build a new thing or improve a process? That’s smart.
Feeling like you have to stuff AI into every new idea or product?
Not so much.