Been thinking a lot about the new patterns of work, and how layering in automation and AI agents can make work both more interesting…and quite frankly, more irritating. 
@Kike_Pena has a new article, You’re missing the point; this is your real value in tech companies, in which he dives into the balancing act of creating great work.
I’m always a sucker for a good infographic as well…
He breaks it down like this with a Thermomix example, you have three choices:
1. Let the machine do everything.
You’ll make food quickly and cheaply, but every restaurant can do the same thing. Your business has no personality, no unique style, and no reason for customers to choose you over anyone else.
2. Do everything by hand.
Your food may be creative and original, but you’ll be slower, more expensive, and harder to scale.
3. Combine the machine with human expertise.
Let the machine handle the repetitive work while you focus on ideas, strategy, customer experience, and creating something unique.
Great stuff, you should check out his article.
Also came across this video today from Natalia Quintero and Dan Shipper, in which she discusses her consulting workflow using an agent to get work done.
It’s a nice overlap with Kike’s point. They really aren’t talking about replacing people with AI. It’s more about building systems in which AI handles execution while humans provide context, judgment, and direction.
I particularly like Dan’s idea that knowledge work is shifting from sculpting to gardening. Instead of creating every output by hand, we’re creating the conditions for good work to grow. That feels very aligned with the Thermomix analogy.
Both pieces point to the same idea that AI is making execution cheaper. And without trying to be trite, judgment is where your attention should be. Whether you call it strategy, criteria, shared context, or decision systems, the differentiation isn’t really the model. It’s how we shape and guide it (ahemm…Glare plug)
I’ve been paying more attention to watch the patterns for patterns in how people use these tools. Progressively, I think we’ll spend less time producing outputs and more time designing the loops that create better ones.
