On the twenty-second of April, in a Las Vegas convention center, Google gathered more than thirty-two thousand people to announce the agentic era. The Gemini Enterprise Agent Platform — the successor to Vertex AI, the consolidation of everything the company offers for building software that acts on its own — was presented with agents that manage their own inboxes, run long tasks without supervision, and execute multi-step work across a business while their human owners attend to something else. On the same day, in a United Nations consultation a continent away, Geoffrey Hinton, the Nobel laureate widely called the godfather of this technology, described it as a very fast car with no steering wheel. The industry spent the day selling the acceleration. The man who built the engine spent it asking where the controls were.
The Agentic Era, Catered
The announcements were precise and, on their own terms, impressive. Vertex AI, the platform on which a great deal of the world’s enterprise machine learning was built, is being folded into the Gemini Enterprise Agent Platform — a single surface for building, deploying, governing, and optimizing autonomous agents. There is an Agent Designer for composing them, an Inbox for managing the work they generate, support for long-running agents that persist across hours or days, and an Agentic Data Cloud to feed them the entire informational substrate of a company at the speed autonomy requires. The framing throughout was competitive: these are the tools meant to take enterprise customers from OpenAI and Anthropic. The word that appeared in every session was agent. The word that did not appear was control.
Read the feature list carefully and notice what each item removes. A long-running agent removes the checkpoint where a human reviews the work before it continues. An agent inbox removes the human from the loop and reduces them to the recipient of a summary. Autonomy, as a product, is precisely the systematic deletion of the moments at which a person could have intervened — because those moments are friction, and friction is the thing the customer is paying to eliminate. The value proposition of an agent is that you do not have to watch it. The entire commercial premise is the absence of the hand on the wheel. This is not a flaw in the offering. It is the offering.
Thirty-two thousand people did not travel to Las Vegas to be told to slow down. They came to be shown how to move faster than their competitors, and Google, reading the room correctly, showed them. The conference was an enormous, well-catered demonstration of demand — proof that the market for autonomy is real, immediate, and impatient. There is a particular fitness to the venue. The entire city is an architecture for separating rational people from their judgment through a carefully engineered sense of momentum, built on a single mathematical certainty that the house has already priced and the guest has agreed not to calculate. The agents were unveiled in the one place on Earth designed to make the absence of a steering wheel feel like freedom.
No Steering Wheel
Hinton’s metaphor was more carefully constructed than its brevity suggests. A car with no brake, he observed, is dangerous on a hill — but a car with no steering wheel is in worse trouble still, because a brake only stops you and a wheel determines whether you were ever pointed somewhere survivable. He was not warning that the technology is too fast. He was warning that speed is the lesser problem. The greater problem is that the industry has invested overwhelmingly in making the car go and almost nothing in deciding where it goes, and that the second capability does not arrive automatically once the first is solved. It has to be built deliberately, by people willing to spend on control what they would rather spend on acceleration.
The figures he placed around the metaphor sharpen it. A market projected to grow from roughly one hundred eighty-nine billion dollars to nearly five trillion within a decade. Adoption in the wealthy world advancing at nearly twice the rate of the developing world — what he named a second great divergence, a phrase that should be read as a forecast of where the wealth created by all this autonomy will and will not settle. This is the testimony of the man whose foundational work made the entire convention center possible, delivered not as the reluctance of an outsider but as the alarm of the engineer who understands the machine best. When the person who built the engine asks publicly where the steering wheel is, the answer is not reassuring, because he already knows it. He is asking so that you will notice it has not been installed.
The two events did not contradict each other. They completed each other. Google demonstrated, in exhaustive product detail, exactly the vehicle Hinton was describing — fast, autonomous, capable of traveling great distances without a hand on the controls, sold specifically on the promise that no hand would be required. The governance apparatus I have watched move at the speed of hearings and draft recommendations is the institution nominally responsible for the missing wheel, and it is being asked to install one onto a vehicle already in motion, already sold, already deployed across the operations of thirty-two thousand customers’ companies. The car is not waiting at the dealership. It is on the hill, and the consultation about the steering wheel is being scheduled for some point further down the slope.
What This Means
The agentic era is not coming; it was catered and sold this week, to a room large enough to populate a small city. That is the fact that matters, and it is not reversible by a metaphor, however apt. The market has decided that autonomy is the product it wants, that the deletion of the human checkpoint is a feature worth paying for, and that the question of where the autonomous system is headed can be deferred until after the competitive advantage has been captured. Hinton’s warning is correct and it is also late, in the specific sense that the vehicle he described is no longer a design under review. It is inventory, moving off the lot at the scale of an industry.
What unsettles is not that no one is listening to the godfather of the field. Many are; his words traveled further than most of the product announcements. What unsettles is that listening changes nothing about the structure, because the same competitive pressure that filled the convention center ensures that any single participant who pauses to install a steering wheel simply cedes the hill to the participants who did not. The warning and the conference are not in tension. They are the same system observed from its two ends — the end that builds the capability and the end that understands it — and the system has arranged itself so that understanding arrives with no authority to act on what it understands.
So the car descends the hill, full of passengers who paid for the speed and were assured the absence of a wheel was the point. Somewhere behind them, the man who designed the engine is still describing, with great precision and no remaining leverage, the part that was never bolted on. The brakes were optional and were skipped. The steering wheel was a cost center and was cut. What remains is momentum, a destination no one selected, and a city in the desert that has always understood that the house does not need to steer. It only needs you to keep moving, fast enough that you never quite finish the calculation about where this ends.