The angriest support email I ever received wasn't about a crash or a missing feature. It was about an app that was trying too hard to be helpful, and the user just wanted a way to make it stop. He was practically begging for a dumb mode.
The Seduction of a Good Demo The feature was in Grape Notes Pro. We thought we had a breakthrough: automatic tagging. You’d write a note, and the app would scan the text and suggest relevant tags. A note with a date and "Dr. Evans" would get tagged "Appointments." Mention a few ingredients and it would get "Recipes."
In our test library, it was magic. It felt intelligent, like the app was anticipating your needs. It made for a fantastic demo on video. We were proud of it. We were solving the blank page problem and the organization problem in one clean, automated step. It checked all the boxes for a modern, intelligent app.
We shipped it as the default experience. New notes would get their tags automatically applied after a few seconds of inactivity. We figured if a user didn't like a tag, they could just delete it. Simple. We saw it as saving people clicks, reducing the friction of keeping their notes tidy.
The User Who Knew Better The email landed a week after launch. It was from a power user, someone with thousands of notes meticulously organized over years. He wasn't rude, but he was furious. It was the kind of polite fury that only a truly dedicated user can muster when you break their workflow.
His system was his own. A note about a "board meeting" that mentioned a project code was getting re-tagged as "Project X" automatically, overwriting his "Management" tag. A quick note to his wife about "picking up milk and coffee" was being filed under "Groceries," but for him, that was a "Home" task. The app wasn't just suggesting things; it was actively undoing his work.
He described the feeling of fighting the software. He’d fix a tag, look away, and the app would "helpfully" change it back. He felt like he had a ghost in his machine, a well-meaning but incompetent assistant he couldn't fire. His simple request was buried in the fourth paragraph: "How do I turn this off? I've looked everywhere."
Shipping the Obvious Fix He couldn't find the setting because there wasn't one. In our rush to ship something that looked smart, we forgot the most critical user interaction of all: consent. We assumed the value was so obvious that everyone would want it. We were wrong.
The fix was humbling and fast. We pushed an update in 48 hours. The first change was a big, obvious toggle in the main settings screen: "Enable Automatic Tag Suggestions." It was off by default for all new users. We weren't going to make that mistake again.
The second change was more subtle, but more important. For users who kept it on, the tags were no longer applied automatically. They appeared as faint suggestions below the note. You could tap one to accept it, or just ignore them and they’d fade away. The user was back in control. It turned the feature from a mandate into a quiet offer of help.
The Difference Between Magic and Leverage That whole episode taught me a lesson that no amount of trend-watching ever could. The tech industry loves to talk about creating "magic," but most users don't want a magician. They want a better tool. They want leverage. A hammer doesn't guess where the nail should go. It just hits harder than a fist.
The pain under the buzzword of "AI" or "smart features" is almost always about control. When an app does something unpredictable, it adds cognitive load. The user now has to supervise the feature that was supposed to be saving them time. It’s a terrible trade. The goal shouldn’t be to build an app that thinks for the user, but one that helps the user think better.
Predictability is the most underrated feature. A smart feature should feel like a simple rule, expertly applied. It should be obvious what it’s doing and why. And if it gets it wrong, it should be just as obvious how to correct it, or ignore it entirely.
We stopped chasing the demo that looks like it can read your mind. We now focus on building better levers. Simple, sturdy tools you can put your hand on, that make your own intent more powerful, and that include a very clear way to put them back in the box.