AI can build almost any feature I ask for. How do you decide which ones to ask for?
by•
When you build with AI, “can I build this?” stops being the hard part. “Should I build this?” takes its place.
Here’s a small example. I’ve been working on an app that checks whether furniture fits in a room. I got feedback from two strangers on Reddit, and it led to the newest feature: a shopping list that keeps each item’s size next to its price and flags when the total goes over budget.
I would not have prioritized that on my own, and it only exists because someone outside my head pointed at it.
So I’m curious how you all decide what to build next. Do you lean on your own usage, on suggestions from the AI, or on feedback from real people? And has anyone built a feature that looked great and that nobody ended up using?
11 views
Replies
Good to see the shopping list came from real people, Joshua. My two inputs are other people's reviews and my own usage data. Before planning features for my song app I read about 200 public reviews of similar apps, and the most frequent request was "keep the song I like, change one thing". On the other side, the data humbles the features I did build: of 51 song templates, 40 had never been opened, and a gift of free credits offered to 75 people was claimed by 11. So the order I trust: what strangers ask for repeatedly, then what my own numbers say people actually touch, and the AI's suggestions last.