As more and more people are using AI, we are starting to see a pattern emerging. Someone needs to create a design or write a user story, so they use AI to do it for them. The output looks great! And it took no time at all. So they send it over. But when a developer starts to ask questions about how to implement their design or story, they lack the fundamental understanding of the problem to answer the questions.
At Uptech Studio, we use AI to help us do our work, not to do our work for us. The difference matters.
In practice that means we tell AI what we need, ask it questions, and use the answers to sharpen our own thinking. When it gives us a recommendation, our job is not done. We read it, push back, and decide what we agree with. We need to be able to explain our reasoning to someone on the team who was not in the chat.
We were working on setting up attribution on one of our projects and the implementation was not straightforward because of how people use this product.
So I opened the project in Cursor and asked AI questions from inside the codebase, so the model had the actual project context. Then I put the responses into a Google Doc and reviewed them myself to make sure I fully understood everything and could speak to it in a meeting.
Only then did I share the doc with the developer on the team. They had clarifying questions, which is exactly what should happen. But we arrived at the same place logically, because the reasoning was visible and reviewable, not trapped in a chat transcript nobody else saw. Or lobbed over in a JIRA ticket with no further explanation.
As long as building software remains a team activity, we need shared ways to make sure everybody understands where we are and why we are doing things a certain way.
That does not mean every decision needs a forty-page spec, but it does mean the team can explain the "why" without opening a tool. AI makes it easier than ever to produce artifacts. It does not make understanding optional.
Otherwise, it is like cheating on a homework assignment in high school. You turned something in, maybe even got a decent grade. But you did not learn the material well enough to discuss it when the teacher called on you. The meeting is the class. The AI-generated doc is the copied answer key.
Use AI aggressively. Use it to explore options, stress-test assumptions, and move faster through the tedious parts. But do the work of understanding what it gave you before you put your name on it or ask someone else to build from it.
