38. AI agents are having their "React" moment
A quick note on the mindset shift needed for agentic AI
Over the past few months, I’ve been leading a bunch of projects at work where we’ve built real AI agents. Actual agents that think, decide, and execute.
And there’s one big learning that keeps coming back to me.
When you start building agentic tools, you realise there’s a shift in thinking required. If you’ve been a developer long enough, you probably remember the jQuery to React transition. Back then, we all knew how to manipulate the DOM directly. You could build anything that way. But React forced us to rethink the whole thing: components, state, data flow. It felt weird at first, then it became obvious.
That’s exactly where AI agents are today.
Yes, a lot of things can be done with traditional engineering. APIs, workflows, cron jobs, pipelines - all of that still works, and will always be required. But for a certain category of problems, agents do it better. And that category is only going to grow as LLMs get better.
The tricky part is, you won’t see the “better way” unless you spend time actually building agents. Once you immerse yourself in it, patterns start showing up. Suddenly, you start thinking, “Wait, why am I manually stitching all this logic together? An agent could handle this entire flow.”
Processes have to be reimagined with a new question in mind -
“How would an agent do this?”
Once you put on that lens, your brain rewires itself a bit. You start spotting opportunities everywhere, places where agents can reduce complexity, automate decisions, or create new experiences that weren’t possible before.
That’s why I keep saying: AI is having its React moment. It’s the same mental leap. Same initial discomfort. Same “oh wow, this is a different way of building software” realisation.
If you haven’t tried building an agent yet, just start.
It’s not as hard as it looks. And once it clicks, it opens up a whole new way of thinking.
Go build one today.