Attention · Absorption · The body
Felt at the Tip
A tool joins your body when its feedback is fast and faithful. The AI that wins will not be the smartest one. It will be the one you stop feeling.
You stop feeling the tool and start feeling the world at its tip.

60 ms. The rod is part of you. The field of “near me” runs all the way to the tip.
You back into a tight space beside a concrete post and you know, without looking, that the bumper has about a foot to spare. You are not feeling the steering wheel. You are feeling the post.
That small miracle is the most important thing I know about tools, and it is about to become the most important thing to know about AI. The car is not in your attention. The post is. Somewhere between your hands and the bumper, the machine stopped being a thing you operate and became a part of you that touches the world.
It happens with a pen, where you feel the grain of the paper at the nib and not the barrel against your fingers. It happens with a fork, a hammer, a keyboard, a bicycle. It happens most famously with a cane. The philosopher Maurice Merleau-Ponty described a blind man whose stick had stopped being an object for him. He did not feel the stick pressing his palm. He felt the curb.
The map of where you end
In 1911 two British neurologists, Henry Head and Gordon Holmes, gave the phenomenon a name: the body schema. It is the brain’s running model of where your body is and where it ends. It is not your skin. They noticed that it could reach past the skin, and their example was charming: a woman’s sense of where she was could extend to the feather in her hat.
Eighty-five years later, a team in Japan led by Atsushi Iriki watched it happen in single neurons. Macaque monkeys have cells that fire when something comes near the hand, as if guarding a little bubble of space around it. The team taught monkeys to rake food toward themselves. After a few minutes of active raking, those same cells began firing for things near the tip of the rake. The bubble had stretched along the tool. When the monkeys simply held the rake without using it, the bubble shrank back to the hand.
The rake did not become smarter. It became theirs, because it answered every motion right away, the same way, every time.
Michael Polanyi, the chemist turned philosopher, put it best: we attend from the probe to the thing at its far end. The tool moves from the object of attention to the instrument of it. That is the whole trick.
- ActiveYou act through it. A rake you only hold never joins you.
- FastThe answer comes back inside the moment. Late feedback reads as someone else’s hand.
- FaithfulThe same motion gives the same result. Surprise from the tool keeps it outside you.
The last condition is where this connects to the rest of my work. A tool you have absorbed is one whose behavior sits entirely inside your expectation. Its Actual equals its Expectation, so its share of Reality is one, and the surprise it contributes is the logarithm of one, which is zero.
Rtool = A / E = 1 → Stool = ln 1 = 0
That is what it means for a tool to disappear. Not that it stops existing, but that all of the surprise moves off the tool and onto the world at its tip. Every jolt you feel is about the post, the curb, the paper. None of it is about the car, the cane or the pen. I wrote in Zero Surprise that whole economies are built to land exactly on expectation. Incorporation is the same law at the scale of one hand.
The turn: AI lengthens the stick
Every tool in history extended the body by a little. A rake gave a monkey an extra arm’s length. A car gives you a lane. A personal AI that acts for you extends you into places no stick has ever reached: your inbox, your calendar, the clinic’s scheduling line, the vendor portal, the county’s permit office.
Most people are judging these systems the way you judge a job candidate. How smart is it, what does it know, can it pass the exam. The body schema does not care about any of that. It asks three things. Do I act through it? Does it answer right away? Does it answer the same way every time?
There are two consequences most people will miss. The first is that the AI that becomes part of you will be the most consistent one, not the most capable one. The second is harder. Once a tool is part of you, you stop seeing it. Martin Heidegger observed that you only notice a hammer as a hammer when it breaks. Until then you see the nail. An incorporated delegate is a delegate you have stopped checking, and that is both the gift and the risk.
In You Cannot Tickle Yourself I argued that acts an AI makes for you land at full size, because your brain did not predict them. Incorporation is the other end of that road. A delegate starts as something that happens to you, full of small jolts. If it is fast and faithful long enough, its acts become predicted like your own, and the jolts move outward to the world.
You are pouring coffee when you know that Thursday’s meeting has moved to Friday. You did not read a message about it. You know it the way you know a stair is shorter than it looked.
Five predictions
Consistency will beat capability in personal AI.
People will keep the delegate that behaves the same way week after week over a smarter one that keeps changing. A model update that alters how your delegate acts will feel like waking up with someone else’s hand, and the best providers will promise sameness across updates the way carmakers promise a familiar steering feel.
Speed will become a matter of the body, not a spec sheet.
For small actions there is a felt line. Below it, the delegate is you. Above it, it is a service you wait on. Companies will stop advertising raw speed and start designing for that line on the hundred little acts that make up a day.
Outages will be described in the first person.
Today people say the app is down. Soon they will say “I can’t do anything this morning,” the way you talk about a numb hand. Losing an absorbed delegate will be treated with the seriousness of an injury, and backup delegates that act the same way will become ordinary.
Good oversight will be designed as a deliberate break.
Because you stop seeing what you have absorbed, the honest delegate will make itself visible on purpose at set moments: here is what I did for you, here is what I chose not to do. Heidegger’s broken hammer, scheduled. The delegate that never surfaces will be the one that drifts unnoticed.
Switching will feel like learning to write with your other hand.
Loyalty to an AI will be motor, not contractual. People will not stay because of a subscription. They will stay because changing would make them clumsy for a month. The provider that wins is the one you can no longer feel.
Where I am living it
I have written that I now tell my AI “you decide.” What I did not see then is that the instruction only works with a delegate I have stopped feeling. I cannot say “you decide” to something that surprises me about itself. I can only say it to something that has gone quiet, the way the cane goes quiet, so that what reaches me is the world. That is what I meant by absorption, and it is why I think the race in AI is being scored on the wrong axis. I wrote in Delegate It that AI lowers the cost of doing. The body adds a condition: doing only becomes yours when it comes back fast and faithful.
The claim ledger
- Already true
- The body schema reaches into tools that are active, fast and faithful: canes, cars, pens, rakes. People already describe their phones as part of themselves, and feel lost without them.
- What has to happen
- Delegates must behave the same way across updates, answer small acts inside the moment, and report outcomes back to the person, so the loop closes on the person and not on a dashboard nobody reads.
- Where I am probably wrong
- Incorporation may be tied to touch and movement. If the brain only absorbs tools that close a sensory loop with the hand, a delegate acting in a calendar will never become part of you. It will stay a trusted colleague, and then the right frame is trust and accountability, not the body, and consistency matters less than I claim.
Back to the post in the parking garage. You were not feeling the wheel. You were feeling the post. The next tools will reach much farther than a bumper, into every office that keeps your life running. The best of them you will not feel at all.
You will feel the world they touch.
Background
- Head, H. and Holmes, G. (1911). “Sensory disturbances from cerebral lesions.” Brain 34. doi.org/10.1093/brain/34.2-3.102
- Iriki, A., Tanaka, M. and Iwamura, Y. (1996). “Coding of modified body schema during tool use by macaque postcentral neurones.” NeuroReport 7. doi.org/10.1097/00001756-199610020-00010
- Botvinick, M. and Cohen, J. (1998). “Rubber hands ‘feel’ touch that eyes see.” Nature 391. doi.org/10.1038/35784
- Merleau-Ponty, M. (1945). Phenomenology of Perception.
- Polanyi, M. (1966). The Tacit Dimension.
- Heidegger, M. (1927). Being and Time.
- Rector, J. The Reality Equation.
- Rector, J. Absorption: A Personal AI and the Work You Can Stop Carrying.