Smarter use of my AI agents: quick wins that free hours
I made smarter use of my AI agents today, so had time to head to the beach for the morning run. Reminded me of more examples of how folks are freeing up hours with AI.
👉 I changed how I fix UI problems in my apps. I used to waste time spelling out all the visual quirks — “the nav bar is 7px too tall on mobile”, “the icons aren’t aligned with the text”, etc. I should have realised earlier the power of applying AI models to this kind of visual detail.
Now I just screenshot the app, upload the image and say “spot what’s wrong and plan the fixes.” Claude Sonnet 4 and OpenAI o3 both do well with this. Faster resolution, with fewer rounds of iteration.
👉 Last week at the Cursor Sydney event, Kevin Wu Won showed how he’s using an agent with the Playwright MCP to automate UX testing. It generates UX tests for a pre-prod build, then runs them on each new release.
I’ve leaned on AI a lot for writing unit tests, but not much for UX — I’m excited to give this a try. I know many QA teams who’d kill for this kind of regression testing on autopilot.
👉 Dan Fowlie at Vhybe Apps takes a friendlier approach to his AI agents. They each get names, desks, keyboards, even chairs. So they work extra hard for him.
More seriously, I love this approach to keeping the human on top of the AI detail as it plays out. And think of the step count from moving between those workstations all day.
The pace we’re learning in this agentic AI space is nuts, let’s see what next week brings.
Bonus points if you can guess the beach. 😎
Originally shared on LinkedIn