please make me think
use the robot to help you think better, not think less
When I started as a product manager, I read a book called Don’t Make Me Think.
I learned how fundamental this book was to tech. Its basic premise: make it as easy as possible for users to complete a flow. They shouldn’t have to think. This thinking is baked into the apps you and I use today. Amazon Prime’s one tap purchase. TikTok’s feed. We barely think when we use our phone.
I see this same principle baked into the way AI tools are designed. Everything is a chatbot. Everything. That’s because we all know how to interact in chat.
Claude Code is great. I use it every single workday. It feels like play to me.
I’ve used GenAI to diagnose car troubles, dog sicknesses, strategize my savings goals. You name it.
Where I struggle is when I’ve used it to skip the part where I think.
Not only is the work not getting the best part of me. But I feel disconnected from the output itself. I just write some words, receive a result, put that in a doc somewhere, and then it’s... done? That’s it!
Behzod wrote recently: Be romantic. AI isn’t that romantic. It wants us to go straight to the chapel before even having a first date. Where’s the fun in that?
Conversations with coworkers and friends revolve around a similar theme. It speeds up work tremendously, and can even do some complex things. But when we use it to do the thinking, we’re less convicted in the result.
I’m not criticizing the immense progress in generative AI. The tools are amazing. I just think we should be using them to help us think better, rather than think less.
John Cutler had a really great quote: Diagnose, before automate. That’s sort of my new mantra. I feel like everyone is using it to automate automate automate. We gotta save time. For what? What do we do with our newfound time?
Everything above can be sharpened by asking three questions about any behavior you’re considering handing to AI. Is it a capability problem? Is it an opportunity problem? Or is it a motivation problem?
Using AI to make a dashboard won’t work when the humans have to change their behavior to keep their dashboard up to date. Behavior change still requires… humans.
Diagnose, before automate. Why are we using this tool? What do we hope to achieve as a result?
That’s the exact part that requires your brain and my brain. You can use the LLM as a thought partner here, but I wouldn’t offload this part to the robot. The thinking needs you.
There is good, difficult, strategic work to do that requires your brain and intuition. Building trust, navigating a tricky power dynamic, fostering psychological safety, encouraging and trialing a new way of working, working towards the same thing, navigating failure and learning together. I’m certainly biased, but those things are really hard to accomplish with a Claude Skill. They help, but ultimately we’re doing this work.
Taste is the new buzzword. OpenAI hires for it. Paul Graham says it’s your main differentiator. But a 2024 study showed that Generative AI enhances our individual creativity, yet reduces the collective diversity of novel content. (Thank you to my coworker Cara Moyer for sharing this study.)
I suspect everyone will develop their individual taste, but collectively… everything will continue to look the same. In a similar way tech bros wear black tshirts, corporate memphis became the norm, and every app became a chatbot.
Your taste, judgment, and creativity comes from you being a human. Trying something that didn’t exactly succeed, being inspired outside or in a museum or on top of a mountain or in a conversation with a friend. Living your best life boo boo.
As generative AI continues to blow our minds, the future becomes more uncertain. Friends of mine looking for a tech job feel hopeless about finding their next. Friends of mine in tech question their job security. There is this lightweight stress going around that no amount of taste or prompting can solve.
I’m not really sure where this will lead to, but I’m going to continue thinking. Please make me think.


Judgement, taste and feeling can be imitated but not replicated by AI. A good reminder, for those of us in the tech almost daily, and hearing the heavy emphasis on using AI for practically everything.