
I'm consulting for a team of about 20 developers right now. Most of them I wouldn't want on my own team.

And everyone above them believes AI will save the team. Give them Copilot, give them Claude, and suddenly everything ships faster.

That's not what I see happening. AI is a multiplier. It multiplies what's already there.

If someone is skilled, reliable and fast, AI takes them to another level. If someone is average, needs hand-holding, and ships features that come back three times for fixes, AI won't fix that. The broken thing just arrives faster.

And what I actually see with these people is something worse. They use AI to make their own work easier, not bigger. Same output, finished sooner, call it a day. So the company now pays for the AI on top of the salary and gets exactly what it was getting before. The multiplier works, it just multiplies their comfort instead of their output.

Code output isn't even the metric that matters. How many meetings do you need before they start working on something? How many times does the feature come back broken? How much micromanagement does the person need? Those are the real metrics, and AI doesn't move them.

So my math is simple. I'd rather pay one person $15,000 a month, a real professional, than three people at $5,000 to $7,000 each. That one person with AI will outperform all three combined. The three average ones can't get there, no matter what tools you hand them.

This is an observation, not advice. But after watching a 20-person team adopt AI, and after talking with friends from other companies who see exactly the same thing, I'm convinced of one thing.

AI doesn't set the ceiling. You do. Your skills, your reliability, your work ethic. That's exactly what AI works with.

To be fair, there's a flip side. It's pretty naive to expect people on a fixed salary to be motivated to deliver twice the output. And whether we like it or not, working this way is mentally much more demanding. AI delivers the things, but the context switching never stops.

That's a different question though, for a different time and a different post.


I'm consulting for a team of about 20 developers, and most of them I wouldn't want on my own team. Everyone thinks AI will save them. It won't. It multiplies what's already there.

AI won't fix your average team


My designer sends me work and I reply five days later, sometimes week(s) later.

Not because I don't care. Because I always feel like I have more important things to do. I want to code, build features, push the product forward. From where I sit, I have the biggest impact when I'm building.

But the math doesn't work.

I pay the people who help me with [Edicek](https://edicek.com): a designer, a marketer, a video editor. Experts are expensive, usually well above the market average. And when one of them can't continue because I haven't replied, it's not just blocking them. It's blocking the entire product, because everything is connected.

So the rule I keep reminding myself of is simple. When someone asks me for feedback, that is the most valuable thing I can do right now. Start the day by unblocking the people who depend on me, then do my own work.

The same thing works in the other direction, when you're the one being paid. Fast responses, fast delivery. Nobody ever told me directly that my quick replies won them over, but people see what the market standard is. It's the cheapest way to stand out and look worth your rate.

And there's a pattern I catch myself in more than any other. When I sit on a reply, it's usually not a time problem. It's that I'm not happy with the result. I don't want to deal with it right now, I don't want the confrontation, I don't want to dig into the fact that what I expected isn't what I got. So I move to work that makes more sense to me.

Because when I'm happy with the work, I reply immediately. Same when I can see we're at least heading in the right direction.

A slow reply is rarely about the calendar. It's a verdict I haven't said out loud yet. And if it keeps happening with the same person, that's the real signal: they're probably not right for the project.

I'm not writing this as someone who solved it. I still reply late. This is a habit I'm only starting to build now, because nobody ever told me about it. Nobody warned me that this is part of the job. So I'm fighting it on my own, starting now.

At least I know exactly what it costs.


I pay a designer, a marketer and a video editor to help with Edicek. Then I make them wait five days for feedback because I'd rather code. The math doesn't work.

Shorten your feedback loop


Twenty years in B2B development and I never once needed to know what users do in my app. Someone in management buys the software, and people use it because they have to.

Then I built [Edicek](https://edicek.com), a B2C product. For the first time, nobody is forced to use anything.

So three months ago I plugged in [LogSnag](https://logsnag.com), a tool that tracks events in your app. You log what users do, and it shows you when they registered, what actions they took, how often they come back, whether you're losing them.

What I saw was brutal.

90 percent of people who sign up in Edicek open the chat, type one message, and leave. Maybe they reply once to the answer, and that's it. They close the app and never open it again.

Part of this is fair. The mobile app has zero onboarding. I wanted to get it on the App Store, so I didn't prioritize it. You open the app and you have no idea what's expected of you. I knew that was bad. I just didn't know where exactly people stop.

Without the data, I would have guessed. Maybe the home screen is confusing. Maybe the save flow is broken. It's neither. It's the onboarding. People get into the app, have no idea what to do, so they type a message into the chat and leave. They never create a card. They never save a link or create a note, which is the core action the whole app is built around.

I don't have the solution designed yet, and that's fine. The point is that I now know exactly where to focus and what to iterate on. Not the home screen, but the save flow. The first minutes of a new user's life in the app.

One honest note on the tool. I don't use LogSnag anymore. I have enough data today and I know exactly what I need, so it was easier to build a custom solution. But if you've never tracked your users and you want a simple high-level view — how the app is doing, how many users you have, how they behave, how active they are — it's absolutely the best place to start.

Users are individuals, but as a group they follow one common pattern. You can't see that pattern from inside the app, and you won't find it in support emails.

Without data you're just guessing. With data you know exactly what's killing your retention, and what to fix next.


Twenty years in B2B and I never needed to know what users do in my app. Then I plugged LogSnag into Edicek and found out exactly where people stop. It wasn't where I would have guessed.

Without data you're just guessing


I tried to build an iOS app five times over ten years. Failed every time. This year I learned Swift in three weeks with Claude Code and shipped it. Apple rejected it four times over in-app purchases, but it's on the App Store. I've written about that one before.

The new story is a Chrome extension.

I always wanted a real extension for [Edicek](https://edicek.com). What we had instead was a bookmarklet: you drag it into your address bar and it works like a button. It did the job, but it was a workaround. And for some reason I was scared of building a proper extension. I honestly don't know why. I just never touched it.

This time I described what I needed, planned it with Claude Code, and forgot about the window. I went back to my day job. When I came back, it was done. I made about five small tweaks, submitted it to Google, and it was approved in one day.

A thing I avoided for years took one day. Most of it happened while I was doing something else.

By April I had two things I'd never built before, both done, both publicly available. Since then the Safari extension shipped too. So did a native iPad app and a Mac app. Every single one of those was unthinkable for me a year ago.

I'm not saying AI writes everything for me. The iOS app was real work, three weeks of learning Swift. But the entry barrier, the part that kept me out for ten years — that's exactly what AI collapsed for me.

Things that seemed impossible might not be impossible anymore.

Try them again. Just try a different strategy this time.


I avoided building a browser extension for years without knowing why. Then I planned it with Claude Code, went back to my day job, and when I returned to the window it was done. Approved by Google in one day.

Retry the things you gave up on


A random guy on Twitter sent me a DM. He found a bug in [Edicek](https://edicek.com): he couldn't log in. I fixed something that day and thought it was done.

Three months later, another DM. Different person, same problem. And only then did I find out what was actually going on.

Sign-in with Apple worked fine. Sign-in with Google didn't, but only for a small group of people: the ones with a privacy setting that says don't share my first and last name with the app I'm signing into. And my code had a check that said no first name and last name, no registration. It threw an error and never finished.

It had been broken like that for over a year. It worked fine for me and for most people, which makes it one of those bugs that's almost impossible to find on your own. I have no idea how many people I lost because they simply couldn't get into the app.

And the detail that still stings: I was paying for users through Apple App Store ads. Paying real money to bring people to a door that wouldn't open for some of them.

The only reason I found out at all was that my DMs were open. Two different strangers could each reach me in ten seconds. If my messages were closed, that bug would still be quietly burning my ad budget today.

That's why it surprises me when I see founders and brands with their messages set to private. Someone sees your video and thinks "I have a job for this person" or "I want to work with this guy." They tap your profile and they can't write you a thing. Gone.

If you're building anything public, go to your settings right now and switch your messages to open. Anyone should be able to reach you.

You're here to be found, not to hide.


A stranger on Twitter DMed me about a login bug in Edicek. Google sign-in had been silently broken for over a year for a small group of users, while I was paying App Store ads to bring them in.

Open your DMs


Someone I know is building a fitness app. The model is simple: the trainer pays the subscription, and all their clients get the app for free.

I have a personal trainer who works with a bunch of people. So I told him — if you want, I'll give my trainer a QR code and he can try it with his clients.

And I could feel it. He didn't want it. He didn't jump at it. He just kind of moved on.

I'm not going to guess why, because I honestly don't know. But I get it. I used to be exactly the same. Someone would offer to share my product and I'd say "cool, thanks" and do nothing. Fear, awkwardness, whatever it was.

At some point I made a shift. Now when someone says their friend might like what I'm building, I immediately ask: what's their name? I write it down. I ask what they struggle with. I try to understand what they'd need from the product.

My wife shared [Edicek](https://edicek.com) in a group of moms she knows. Just a simple intro, what it does, why she likes it. Many of them signed up, and today they're some of my most active users. The credibility was already there, because someone they trust said "check this out." No ad can do that.

It's strange when you think about it. Someone offers to recommend the thing you're building, completely for free, and you let it pass. Free distribution, free feedback, free credibility, and the only price is asking one question.

So next time someone offers to share your product with a friend, don't just say thanks. Ask for the name.


I offered to bring a fitness app founder his first real users through my personal trainer. He didn't jump at it. I used to do the same thing, and it's the most expensive mistake you can make for free.

Stop letting warm leads slip


The company I'm consulting for has about 20 developers, and they want to integrate AI into everything. Sounds smart, right?

But every feature they ship has to travel through 20 repositories. Approval processes, code reviews, deployment pipelines. A simple CRUD, creating, updating, deleting and listing tasks, takes them a day and a half.

It hit me the day after our meeting, so I went back and proposed a challenge. We agree on one feature. I build it in my system, they build it in theirs. My estimate for my side: an hour, maybe two. Their side: the day and a half. They agreed, and immediately started asking how I'd approach it.

The point of the challenge isn't that I'm faster. It's that building the same thing in both systems makes the bottlenecks visible. It shows the target state: how simple and fast the same work could be.

Because AI doesn't fix bottlenecks. If your basic task takes four times longer, you need four times the people. That's the whole math. My team of 5 can compete with their team of 20, not because we're smarter, but because we don't have 20 repos and 6 approval steps slowing every single change down.

When the basics are this slow, adding AI just means the requests pile up in front of the same gates.

So before you integrate AI, before you hire more people, before you add new processes: find what's actually slowing you down.

That's where you start.


The company I consult for has 20 developers and wants AI in everything. But every feature crosses 20 repositories and 6 approval steps. So I proposed a challenge: same feature, my system vs theirs.

Petr Homoky.

Fix bottlenecks before AI.

Straight to your inbox.

AI won't fix your average team.

Shorten your feedback loop.

Without data you're just guessing.

Retry the things you gave up on.

Open your DMs.

Stop letting warm leads slip.