You shouldn't multitask, even with AI
May 25, 2026
Watch on YouTubeHey everyone, this is Allan and today I'm going to talk a little bit about trying to do a lot of things at the same time and mainly using AI.
Using AI
I've been working with AI a lot, not only with cloud code in programming but also since ChatGPT launched. I was always using it to work and to study. In February I started using cloud code more seriously and realized that, in theory, I could do a lot more things at once. I thought that because AI can do things for you while you’re doing something else, you could be more productive.
Multitasking Myth
After a couple of months I saw that the idea was wrong. For example, if I try to work on a programming system, plan Instagram posts, organize my second brain, and review topics I’ve studied, all at the same time, my mind gets anxious. I don’t pay attention to anything properly: I don’t review code correctly, I don’t plan good posts, and I don’t understand the content I’m trying to study.
Context Switching
What pushed me to rethink multitasking was the book The One Thing and its discussion of context‑switching costs. When you change the nature of what you’re doing, your mind needs time to refocus. If I start planning an Instagram post and ten minutes later pause to look at code, my mind is still thinking about the post, so I can’t understand the code right away. I end up creating posts slower and planning code slower. When I’m writing instructions for AI, I even listen to music because I need to pay close attention; creating something that people will use, that needs to work and be secure, takes a lot of focus.
Depth vs Speed
Other people who run many agents at the same time can build faster than anyone else, including me. But they also create more depth to review: not only technical depth in the code that needs fixing, but also knowledge they don’t yet understand. Good things take time. Building a product faster doesn’t automatically make more people buy it. If you don’t have good marketing skills, no one will know your product.
Validation Matters
Burning tokens isn’t the same as validating ideas in the real world. It doesn’t matter if you have ten SaaS ideas running in cloud code; the solutions might be bad and no one will want to use them because you haven’t validated them with real people and real problems. That validation phase takes the most time, and it’s impossible to multitask on it.
Focus Needed
The more you try to do at the same time, the more you realize you are the bottleneck. Our attention needs to be used as effectively as possible. With that restriction, we need to focus and not multitask; we need to try to do less, just as I said in yesterday’s video.
AI is a lever, but if you try to push a lot of levers at the same time it reduces the individual force in each one. That’s it for today.