I think that the most affected by AI were:

creating content (text, photos, now video)

creating the code
simplifying the process via AI agents (e.g. agenda)

Someone asked me for help with a problem last month. I gave them a solution. A good one. Thoughtful, detailed, exactly what they had described.

They did not use it. I could tell because the problem was still there weeks later.

I cancelled something last week and I could not explain to the founder why. Not because the reason was complicated. Because it was embarrassing. The product worked fine. I just stopped opening it.

That is the most common kind of churn and almost none of it gets reported honestly. People say price or features or a competitor. The real answer is usually that a habit never formed. The product was fine and it was not quite the shape of my day.

There is a feature in my product that exactly one customer uses. It costs me infrastructure, maintenance, and a little attention every month. The numbers say I should have killed it two years ago.

I keep it because I remember the conversation when they asked for it. They were at a company with a real problem, and I was small enough to actually solve it for them. It was one of the first times I felt like I was building something that mattered to a specific human being.

A person bought my product a year and a half ago, paid once, never upgraded, never asked for anything.

Then they emailed me out of nowhere with a detailed explanation of a problem they had solved with it. Something I had never thought it could do. They had built a whole process around it and wanted to tell me.

I set a rule for myself last month. No new features. No new projects. No starting anything.

Just finish what was already open. Close the loops I had been leaving open for months. Answer the emails I had been avoiding. Fix the small things that had become background noise.

Building the requested feature directly usually leads to bloat and partial solutions.

Asking "What are you going to do with this file/data once you download it?" has saved us weeks of work before. A simple scheduled email or minor tweak often solves the pain point much faster.

How do you document these underlying workflows so the whole team understands the real user intent?

I built two things the same year. One got thousands of visitors in a week. The other got a slow trickle that never really stopped.

I spent all my time on the first one because that is where the attention was. Fed it, fixed it, promoted it. It peaked and faded and now it is a folder I do not open.

Ran into this problem last week: my free plan and paid plan were doing the same thing, just with a lower number on one side. Free users got a report and a tip. Paid users got... the same report and tip, regenerated with a slightly longer prompt. Nobody's upgrading for that, and I don't blame them.

What actually fixed it wasn't adding more features; it was making the paid tier answer a different question entirely. Free tells you what happened. Paid finds out why it keeps happening and gives you a week-long plan to fix the actual root cause, not just today's symptom. Same data, completely different job.

The part I'm still not sure I got right: how far do you take that gap before free starts to feel deliberately crippled instead of just simpler? I don't want it to feel punishing to not pay; I want paid to feel like a genuinely different product built for people who are serious about the outcome, not just "free but slower."

If you've shipped a free/paid split, where's the line between "simpler" and "deliberately withheld"?