We treat friction as something bad. It makes us slow, inefficient, messy. Products brag about removing friction. Entire companies exist just to be “frictionless”.
And mostly, that’s right.
But friction was quietly doing a second job: it made us think before we built.
Before AI, writing code was expensive, so the cost of building imposed a higher threshold on which ideas survived long enough to be built. You had to be selective because the effort itself forced you to decide what was worth making.
AI reduced the cost of getting to a prototype. I can try an idea in minutes, which makes it easier to start building before I’ve decided whether it’s worth pursuing.
This worry isn’t new. Every tool that removed friction got a similar complaint. Practically the same even, if you dive deeper into it.
In the Egyptian myth of Theuth and King Thamus, Socrates tells Phaedrus the story of the Egyptian god Theuth, who invents writing and presents it to King Thamus. Theuth claims writing will make people wiser and improve their memory. Thamus replies that it will do the opposite:
This discovery of yours will create forgetfulness in the learners’ souls, because they will not use their memories; they will trust to the external written characters and not remember of themselves.
”The point is that writing can create the appearance of knowledge without genuine understanding or the ability to think for oneself. It felt cheap to some to just read one’s experience, just like AI-generated things today feel to us.
More than two thousand years later, assembly programmers worried that C would let people program without understanding the machine underneath. Essentially removing the friction of understanding the machine.
If we look at these lessons from history, what Plato said, or what the assembly programmers thought didn’t fully turn out to be true. People read books and get a lot out of them, and lots of programmers write valuable C code without understanding the machine underneath.
I still think friction is a good thing, but you need to be selective about where to apply it.
You don’t want to add unnecessary friction when doing market research, you want the agent to do the research for you. You don’t want to add unnecessary friction when summarizing feedback and categorizing it if you have a ton of it. Instead, you let the agent go through it, categorize all of it, and create a summary for you so you can draw your own conclusions.
You need to add friction where your opinion matters. If you are building something, friction should be added as you write your initial prompt, so that the goal is clear and you know exactly why and what you are building.
One way of adding friction that I like to use is answering questions. Use an agent to bombard you with questions about your idea, or the problem you are facing. This will help you think through it and offer you clarity that you might not have had otherwise.
I even made a skill for it, you explain your idea or problem, and the skill will ask you lots of questions to help you. You can get it on the skills page.
/emil-product-thinking a Chrome extension that summarizes any YouTube video into bullet points, freemium, $5/month for unlimited
This skill is opinionated in that it asks you questions based on the notes I’ve made from podcasts, books, and other sources. So the knowledge and judgment come from there rather than from the agent.
If you want a more general “ask me anything” skill, you can use Matt’s grill-me skill which does roughly the same, just in a less opinionated way.