💯 AI Wrapped: 5 things that changed about work in 2025

Every year has its loud stories and 2025 had plenty of those:

• new models (so many models)

• bigger benchmarks

• faster outputs

• hotter takes

• Sam Altman turning water into Ghibli art

But looking back, the most interesting changes weren’t loud at all. They were subtle, repetitive, and easy to miss in the moment.

So before we all log off for the holidays, here’s a quick AI Wrapped of the patterns that kept showing up in our work all year.

1. The real barrier was never the tools

Here's something I didn't expect when I started reading your emails this year. Nobody was asking "which AI tool is best?" They were asking things like:

"I can't crack the code on how to prompt for HIGH QUALITY outputs."
"How can I trust it when I don't know better?"
"I'm tired of teaching the same tone to the same AI model over and over again."

When we surveyed thousands of professionals taking our 30 Days of AI challenge, 80% of them said the biggest thing they gained was confidence. Which means most people aren't stuck because AI is hard but because they think everyone else already figured it out (they didn't).

2. The prep work ate the prompt

The people getting 10x results from AI weren't better prompters. They were better thinkers.

McKinsey's 2025 report had this stat that's been living in my head rent-free for a while: 92% of executives say they're investing more in AI. Only 1% say their teams are actually good at using it.

That's a 91% gap between buying the gym membership and actually showing up. What closes the gap is the magic that happens before you even open ChatGPT:

• What problem am I actually trying to solve?

• What context does the AI need to help me?

• Where does my judgment still matter?

Those are more “thinking” questions than prompting questions.

3. You don't learn AI. You become someone who uses it.

This one took me a while to articulate.

We ran multiple rounds of the 30 Days of AI challenge this year. Thousands of people. And in the final reflections, nobody said "I learned X tool." They said things like:

"I didn't expect AI to change the way I think, not just what I can create."
"I'm less afraid to experiment now."

That's identity, not information. It's why people are now adding "Vibe Coding" as a skill on their LinkedIn profiles. We're watching real-time as professional validation shifts from credentials to proof of work. From "I know this" to "I built this."

Harold, our bootcamp lead, walked into my first cohort four years ago as a paying non-technical student. This year, he ran 22 cohorts with 600 students and 186 live sessions. That loop, from learner to doer to teacher, is the entire game.

4. Everyone's learning alone (and it's holding them back)

Here's the uncomfortable part.

90% of employees use AI in their personal lives. Only 40% have access through official company channels. So everyone's out here solo-learning, guessing, copy-pasting prompts from strangers, and calling it strategy.

The teams pulling ahead aren't the ones with the best tools. They're the ones experimenting out loud: sharing wins, sharing failures, building collective muscle memory.

(Speaking of learning stuff together, we've been dropping weekly breakdowns on LinkedIn for anyone who wants to learn while doom-scrolling. This week's carousel covers the latest research about how we’ve been using AI at work in 2025.)

5. Creation got cheap. Judgment didn't.

This might be the most important shift of the year.

Anthropic interviewed 1,250+ professionals about how they use AI. One finding stuck with me: even inside AI-native companies, most people delegate only 0–20% of their work to LLMs.

Not because AI can't do more. But because knowing what's good is still hard to outsource.

It's why OpenAI's GDPval report found that the winning setup isn't "AI does everything." It's human + AI where AI drafts and humans decide. That combo cut costs by 63% and time by 39%. When everyone can generate, the work itself isn't the differentiator anymore. Your judgment is.

Max 👋

P.S. Want to make your team & company AI-first? Let us help here.