2 min read
Changing my AI Way
A little bit of the history of the past few weeks on my professional touches with AI
First of all, writing is harder than I thought, and keeping it something up to date is even harder. Kudos to everyone that can keep posting consistenly every week.
Now, it’s been a long time since the last post because a LOT has changed in my life professionally. I changed teams whilst still participating in a huge company-effort task that takes up to ~50% of my week. You can imagine how much time I had to put in to both teams so I can keep the bar high haha. But it’s been great, I learned a new lots of things.
My workflows with AI a few months ago
The first half of the year I could increase my productivity by a good margin basically using AI as brute force companion that I could delegate multiple tasks.
CVEs, architecture implementations, bug fixes, crashes reports, KTLO tasks, you name it. I had literally an orchestrator on my end so it could articulate the implementation with any model and I’d be reviewing, most of, the outputs.
But this is what everyone should be doing right? Incresaing output, making more changes, more implementations, more mesasges, more… everything.
It can get a bit overwhelming and tiring. It is a good experience though, I can’t deny it.
However, I was feeling that I wasn’t going further with AI and got stuck on this orchestrator/reviewer loop where nothing never truly ends and things just keep going faster.
What’s changed
An opportunity inside the company has surged to change to an AI focused role where we can be truly free and go to the limit on this new technology we’re facing now.
Instead of just force bruting output, I now got the time to learn what AI means for how I work with my teammates.
This is just the beggining, but is already so good that I took the time to share.
What’s new
I’ll try to bring more on to that later down the road, but I want to open the eyes of whomever reads it to a better definiton of how agents can work as a system, not just as coding helpers.
Instead of turning them into pure implementation engineers we can create a whole ecossystem where dedicated agents with defined tasks and iterations can help us build the future.
I’m still getting there, but Google ADK will be the basis of these blog posts for a while.