By Toon Timbermont in AI — 01 Feb 2026

Four AI Principles That Keep Me Grounded

After a year of daily AI use, here are four principles that have helped me find my footing. Maybe they'll help you find yours.

Photo by Alternate Skate / Unsplash

Another AI announcement. Another "this changes everything" headline.

I let it go. I'll catch up later.

Don't get me wrong—I'm genuinely intrigued by how fast AI is evolving. It stimulates me. It pulls me in. Over the past year, I've dived deep: experimenting daily, following the tooling, pushing it into my actual work.

And it has changed the way I work. Drastically.

But here's what I've learned: change in the labs doesn't equal change in the field. The breakthroughs in research don't always translate to breakthroughs in my day-to-day.

Moreover, I need to stay grounded. Keep doing the work without being fixated by the fear of becoming obsolete.

So I went looking for something to hold on to. A way to make sense of all this without drowning in it.

I came up with four principles. They're not predictions about where AI is heading. They're my anchors—things I return to when the noise gets loud.

If you've felt that mix of excitement and overwhelm—impressed by what AI can do, yet exhausted by the constant churn—you're not alone.

Maybe these 4 principles will help you too.

1. AI Accelerates 🚀

AI makes things go faster—including things moving in the wrong direction.

The data is compelling. Productboard reports that product teams save an average of 4 hours per task with AI assistance. GitHub's research shows developers complete tasks 55% faster with Copilot.

We are not mature yet in reliably measuring productivity gains, yet my personal experience confirms this. I am way faster in doing product work (or dev work).

But faster isn't better if you're building the wrong thing.

The bottleneck is shifting from "Can we build it?" to "What's actually worth building?" When delivery speed is no longer the constraint, the quality of your product decisions becomes the differentiator.

This means that getting your fundamentals right is more important than ever. Discovery before delivery. Honest assumption validation. Proper development practices. Attention to culture. AI amplifies your direction—make sure it's the right one.

2. AI Augments 🔧

AI broadens your skills—but only if you can guide it and interpret the results.

According to Productboard, 100% of product teams now use AI tools, with 94% using them daily. The opportunity is real: competitive research, presentation creation, prototyping and feature development —all faster.

I find myself doing a broader set of work as well, augmented with AI. Covering both the product side and active development.

The pitfall? Doing everything without understanding what you're doing.

Ethan Mollick calls this the Jagged Frontier. AI can be bad at seemingly easy tasks and excellent at hard tasks. But the line between what's easy or hard for AI is jagged and invisible. You can't tell from the outside which tasks AI will nail or fail.

Only someone with domain expertise can recognise when they're on the wrong side of that frontier—and intervene.

As Adrian Cockcroft (former Netflix) observed: "The skill set needed to use AI agent swarm tooling well is a mixture of product manager and dev team manager."

This means domain expertise becomes MORE valuable, not less. AI can take you from novice to intermediate quickly. Beyond that, you need real expertise to guide and verify.

3. AI Evolves 📈

After massive core model improvements, the toolchain is now improving rapidly.

I follow Ethan Mollick's advice: "Assume this is the worst AI you'll ever use."

We’re seeing several improvement tracks:

Core models keep getting better (though there are indications we're reaching a plateau)
Tooling is starting to mature (MCPs, skills, agent frameworks)

The "how to use AI" is changing as fast as "what AI can do."

Yet McKinsey reports that while 92% of companies are increasing AI investment, only 1% consider themselves "mature." The gap isn't capability—it's learning.

The implication: invest in learning the toolchain, not just the chat interface. MCPs connecting AI to your data. Skills and workflows for repeatable processes. Semi-structured approaches that grow with the technology.

4. AI is Hyped 📉

I find AI amazing. And acknowledge a trough of disillusionment is coming.

Consider the economics. Software companies are suddenly back in the capital-intensive hardware business: data centres requiring water, electricity, and land. GPUs with limited lifetimes and massive depreciation. Oracle's capital expenditure surged from $7 billion to $21 billion in a single year, with plans to hit $35 billion in 2026. This will impact AI pricing.

The adoption numbers tell a similar story. While 100% of PM teams use AI, only 1% of companies are "mature." Developer trust in AI tools dropped from over 70% to 60% between 2023 and 2025.

Remember: AI is excellent for unstructured, ambiguous tasks—natural language, content generation, fuzzy requirements. But for structured problems with clear inputs, outputs, and business rules? Classic computing still gives you deterministic results at a fraction of the cost.

As a company, the key is to build sustainable AI practices, not hype-driven ones. Focus on use cases with clear ROI. Don't over-rely on capabilities that might become more expensive (or stress test the business case). And invest in building human judgment alongside AI capabilities.

The Meta-Insight

These four principles share a common thread: AI doesn't replace your fundamentals—it amplifies them.

If your product thinking is sharp, AI makes you faster and broader. If your foundations are weak, AI helps you build the wrong thing more efficiently.

The PMs who thrive won't be those who use AI the most. They'll be those who know when to use it, when to override it, and how to stay on the right side of the jagged frontier.

Don’t hesitate to reply to this email if you want to dive deeper into how I use AI to be a better product & tech leader!

1. AI Accelerates 🚀

2. AI Augments 🔧

3. AI Evolves 📈

4. AI is Hyped 📉

The Meta-Insight

Further Reading and References 📚

Subscribe to Toon Timbermont