AI/ML
July 27, 2026
It's mid-2026 and the headlines have settled. AI is part of the infrastructure and it's here to stay. Almost every tool you use has it baked in, every team talks about it in their weekly meetings, and it's becoming less and less of a novelty. As teams we're left with a harder question of how do we create value from this?
If you've spent time in product management, you've written a brief, a product spec, a requirements document. You're familiar with the feeling of turning a vague idea into something specific enough for a team to build without ambiguity and with clear success criteria. This is what effective AI prompting is, and it's becoming a defining skill of 2026 because most people still aren't doing it.
You don't need to use elaborate thousand-word prompts or stand up an entire business based on AI like they do in the headlines to create value; you just need the discipline to turn fuzzy thinking into precise input. While those stories are fantastic, most of the value in AI comes from mundane use cases applied consistently rather than spectacular ones.
At Relevate Health, I use AI as a constant part of my workstream while making games for doctors, and not for anything revolutionary, but for the things that fill a typical product managers calendar.
For example, I use AI to draft messages to stakeholders. Not because I can't write, but because I can use AI to sketch the first pass quickly, then spend my time refining tone or tailoring specifics rather than staring at a blank screen. I brainstorm product concepts by asking it to expand on early ideas then filter ruthlessly. I create presentations by feeding it sections of thinking and having it structure them so I can focus on substance rather than format. I build html prototypes to test an interaction pattern, passing requirements the same way I would a developer a brief. I clean datasets by writing prompts that catch edge cases I could manually miss.
None of these are flashy, and none of them require advance prompting however, they all require clarity. The more precisely I define the problem, the context, and the desired outcome, the more useful the response. If I'm vague, I get back mediocre results and need to iterate more heavily, similar to how product requirements work.
This is where I think a lot of the current noise around AI can become a liability. The headlines celebrate the moonshots, which are absolutely real, but those stories can create a feeling of overwhelm for anyone using AI more simply, thinking they're somehow doing it wrong or falling behind when they're not. You don't need to do technically sophisticated work to get value out of AI, you just need to be disciplined about what you're trying to solve, and specific about what success looks like. The best use of AI is often not the most visible one. It's the thoughtful, unglamorous work of reducing friction in your own workflow so you can spend more time thinking and less time doing. Drafting instead of writing from scratch. Exploring instead of deciding prematurely. Testing instead of debating.
That's not falling behind. That's product thinking at scale. And it's the trend that matters most right now.