In partnership with
I was having a chat with a couple of friends about how comfortable we are sharing what we know in public. Two of us work in tech. The third doesn't. She'd just vibe-coded a genuinely useful tool for a client, not a side project she was tinkering with for fun, but something that solved a real problem. She was proud of it. She also wasn't sure she should post about it, write about it, or bring it up in a room full of people who write code for a living.
I know all too well how that feels. Even I catch myself wondering if what I'm sharing is too basic. Some tip about prompting or a workflow that feels obvious inside Zapier, and I hesitate. Then I remember where I was six months ago, and that I'm miles ahead of that version of myself. Plenty of people are just now getting started, and what feels old hat to me might be exactly what they needed to hear.
Part of the hesitation is just putting yourself out there. But part of it is that the baseline for what counts as "using AI well" depends heavily on where you work, and she didn't have a company giving her a script for any of this.
At Zapier, we've been living in this for a while. People are expected to work with agents, use AI in their day-to-day, move faster than they could a year ago. Everyone is being evaluated on it, not just engineers. The minimum bar inside our org is probably higher than at a company that's still figuring out its AI policy.
That gap shows up everywhere once you start paying attention. I've written about resetting engineering expectations before. What's normal here is still a stretch goal somewhere else. Same title, different baseline. And if you're hiring, coaching, or thinking about your next role, you need to know which side of that you're on.
A thank you from this week’s sponsor (I wrote this newsletter using Wispr Flow!):
Talk to your AI tools the way you'd talk to a colleague.
You don't send a colleague a three-word brief. You explain the context, the constraints, what you've already tried. But typing all that into ChatGPT takes forever — so you don't.
Wispr Flow lets you speak your prompts instead. Talk through your thinking naturally and get clean, paste-ready text. No filler words. No cleanup. Just detailed prompts that actually get you useful answers on the first try.
Millions of users worldwide. Works system-wide on Mac, Windows, and iPhone.
What a higher floor actually looks like
On my team, the baseline has shifted. Engineers are closing more PRs. I'm using AI for meeting prep, follow-ups, and the admin work that used to eat my afternoons. PMs on other teams are prototyping faster than they could before. Nobody announced a new policy. The bar just moved.
More PRs can mean more regressions. Or it might not, if you've spent time on async agents that review pull requests and flag problems before they hit production, on top of whatever your CI/CD already catches. From the outside it can look reckless. Sometimes it is. Sometimes the company has just built different guardrails and you can't see them until you're inside.
If you're leading at a company that's already here, your job is to name where the bar is and help people hit it without burning out. If you're at a company that's earlier, your job is messier: make the case, carve out room to experiment, and be straight about what's coming. I've done both kinds of conversations this year. Neither is easy.
Knowing where you fall on the spectrum
Where your company sits changes how you talk to people about AI, not just what you expect of yourself.
When I talk to engineers on my team, most of them are already using it daily. I'm not selling them on the idea. I'm asking what's working, what's still clunky, and where they're stuck. When I talk to a peer at another company who's still fighting for anything beyond Copilot, we're in a different conversation entirely. How do you make the case to leadership? Where can you experiment without getting shut down? How do you get started when the official answer is still "wait"?
Same manager skill. Completely different starting line. Set the bar too high and you'll frustrate people who don't have the tools or the cover to try. Set it too low and you'll have a team that gets surprised when the rest of the industry moved on without them.
A few things I ask when I'm trying to figure out where a company actually is:
Can people pick their own tools, or is there an approved list?
Does AI come up in performance conversations, or is it still treated as a nice-to-have?
Do leaders talk about how they use it, or is everyone figuring it out on the side?
Has anyone said out loud that output expectations changed?
Your answers tell you whether you're setting the floor or still trying to get leadership to agree there should be one.
The other end of the spectrum
Some companies are much more locked down on AI tooling. Copilot only. No access to Claude or Cursor. Strict policies about what can and can't go into a model. Sometimes that's for good reason: security, compliance, data sensitivity. I'm not dismissing that.
But the practical effect is that people in those orgs don't build the same muscle. They're learning one toolchain, one policy, one narrow version of what AI at work looks like. Fine if you're staying put. Rough when you start interviewing.
I've talked to people who only realized the gap when they got on a call with a hiring manager and the questions assumed a year of daily agent use. They'd been doing good work. They just hadn't been allowed to do it that way. The candidate in the next slot might have a year of shipping with agents under their belt. Same job title. Different interview.
Worth finding that out before you're in the room, not after.
When your company won't give you the runway
Here's the part I know is uncomfortable, and I spent some time trying to figure out how to best explain this.
If your company won't give you the tools, the space, or the time to build fluency on the clock, you're stuck with two paths. Neither is great.
Option one: Look for a company that's further along, or at least interviewing like they understand not everyone has had the same runway. Plenty of places are still writing their AI policy. You might fit there fine for now. That window exists. I don't know how long it lasts.
Option two: Build on your own time. Side projects. Writing about what you tried. Anything that gives you something real to point at when a job description assumes you've been living in this stuff for a year.
I know how that lands. "If you want to get ahead, use your free time" is the kind of advice that makes people angry, and often for good reason. Not everyone has free time. Not everyone should have to subsidize their employer's slow adoption with their evenings and weekends. I'm not saying this is fair. I'm saying it's the situation a lot of people are in, and naming it is more useful than pretending the gap doesn't exist.
If you can learn on your own, it opens doors. If you can't, that doesn't make you lazy. It might mean the company you're at isn't going to get you where you want to go, and that's worth weighing honestly.
What I'd take from this
If you're leading: figure out where your company actually is on this spectrum, then coach from there. Don't assume your whole team has the same tools, the same mandate, or the same head start. Say what you expect. Say what you're doing to help people get there. If the bar went up and you haven't said it out loud yet, say it. If you can't give people what they need to meet it, say that too.
If it's just you for now: compare yourself to the market you want to join, not just the people sitting next to you in Slack. If you're at a locked-down company and you might want out eventually, start closing the gap however you can. If you're somewhere that's already far along, check whether you're actually keeping up or just benefiting from the org being ahead of you.
My friend built something real and still wondered if it counted. I second-guess whether my posts are too basic. The gap between companies is only making that worse, because "normal" keeps moving and nobody's posting a single definition of what good looks like. All you can do is know where you are, coach from there, and share what you know even when you're not sure it's enough yet.