I think that in all this talk about AI, we forget one simple thing that Scrum has been giving us for years, and that is the daily stand-up. For a long time, many of us saw it as just another meeting, fifteen minutes that interrupt our work every morning. But since AI became part of our everyday work, I've started to see it differently.
When each developer works with their own AI agent, it's very easy for everyone to go in their own direction. One person gets a suggestion from AI to change how some part of the system works. Someone else spends two days building something based on an assumption that AI simply made up. AI gives answers that sound very confident, and because of that we often accept them without asking too many questions.
I saw this very clearly on one of the projects I worked on, where we didn't have stand-ups at all. Our bosses decided that we didn't need them. The idea was that we were all seniors, we knew what we were doing, and if something came up we could just send a message on Slack.
It sounded like a chill and great idea, but in practice, so much was lost. Most of the team knew their own tasks, but nobody really had the full picture of what we were building and why. It happened regularly that I got assigned to review a PR for a feature I had never even heard of. So I would open it, see dozens of files of AI-generated code, and try to figure out from the code itself what the feature was supposed to do. When I asked something on Slack, I would sometimes get an answer that was clearly written by AI too. So I was reading AI slop code and then AI slop explanations of that code, and in the end I was more confused than when I started.
After a while, it became hard to stay focused. You spend your day jumping between PRs you don't understand, messages that don't help, and your own tasks that you're trying to finish in between. Slack messages are fine for quick questions, but they can't replace the moment when the whole team is in the same place, even for fifteen minutes, and simply talks about what is going on. Being senior doesn't change that. Even the most experienced developer can't review a feature well if nobody told them what it's for.
This is where the stand-up helps. Every day, the whole team hears what everyone is working on and why. If someone went the wrong direction because of something AI suggested, there is a good chance that another person on the team will notice it right away.
The stand-up can also be a good place to check AI ideas with other people. It's completely normal now to say something like "AI suggested we solve this in a certain way, does that make sense to you?" Someone with more experience in that part of the system can tell you in a minute if it's a bad idea. The question about what is blocking us also gets a new meaning. Sometimes the blocker is the AI itself, when it keeps giving wrong or conflicting results and we keep trying new prompts instead of asking a colleague. Saying it out loud in front of the team often solves the problem faster than another hour of prompting.
Fifteen minutes every morning is a small price for a team that actually understands the project it's working on, and that can still fix an urgent bug five minutes before the release, even when the AI agent isn't responding.