Uber owns dispatch software much of this trade runs on. Who owns your passengers?

On 26 March 2026 Uber, Verne and Pony.ai announced a three-way deal to bring robotaxis to Europe. Pony.ai supplies the autonomy, the Croatian company Verne buys the cars and keeps them on the road, Uber brings the network. That same announcement already says riders “will be able to hail one via Uber as well as through Verne’s app”, and that Uber “intends to invest an undisclosed amount into Verne”.

Two weeks later, on 8 April, Europe’s first commercial robotaxi service opened in Zagreb: Verne’s app, ninety square kilometres of the city with the airport inside it, from seven in the morning until nine at night. On 19 August the rides appeared in the Uber app as well, with a licensed operator on board behind the wheel. On 10 September the first rides with nobody at the wheel began, over twenty-two kilometres to the airport: test rides, by invitation, while the commercial service still carries an operator. On 7 October Verne’s chief executive confirmed the investment.

The papers wrote about the car. For anybody running a fleet the thing to look at is different, and it is not an ambush: it is a choice. The most advanced operator in Europe, the one that actually owns the driverless cars and has an app of its own, decided from day one to go through the aggregator’s app as well, and to let it onto its share register.

If that is what a company with robotaxis does, it is worth asking what a firm with the shifts, the licences and thirty years of customers will do, when it has no list to start from.

And if that sounds like somebody else’s problem, the same thing has already happened here, with human drivers, to the software you dispatch on.

Licensing protects the licence. Parliament has already written down what happens to it

The instinct is that regulation holds the line. It does, for the thing it was written to hold: the licence.

Look at what Parliament did with the rest. The Automated Vehicles Act 2024 creates a permit for automated passenger services, and section 83 is titled “disapplication of taxi, private hire vehicle and bus legislation”. Where a permit holder runs one of these services, the Act says the vehicle “is not to be treated for any purpose as being, or as being used or operated as” a hackney carriage, a private hire vehicle, a public service vehicle or a hire car.

That is not a forecast and it is not a reading. It is a section heading. The framework that defines your trade was written with a door in it, and the door is marked.

Read the next section too, because it is the part that is usually left out. Section 85 is headed “Consent requirement for services resembling taxis or private hire vehicles”, and it says the permit may not be granted “without the consent of each licensing authority in whose area the service may be provided”. So the door has a lock, and your licensing authority holds the key.

Then read how the lock works. The authority has six weeks. If it fails to give or refuse consent in that time, consent is deemed given. If it refuses but fails to put its reasons in writing, consent is deemed given as well. A lock that opens when nobody answers the door is still a lock, but it is not a strong one, and six weeks is not long in a licensing office.

What no Act here has done is require a passenger to ring one number rather than another. The customer goes where booking is easiest, and asks nobody’s permission.

The script is already written, and it was written here

You do not need the robotaxi to see how this ends. It has happened twice already, with drivers in the seats.

August 2020. Uber buys Autocab, the company that supplies booking and dispatch technology to private hire operators, which is to say to the firms Uber competes with. On 29 March 2021 the Competition and Markets Authority closes its inquiry having “found no competition concerns as a result of this deal”, and the purchase completes on 14 April.

It is worth reading why it closed. The authority put to itself exactly the data question, whether operators could be forced to pass data to Uber, and set it aside because there are “other credible suppliers of BDT and referral networks that these taxi companies could switch to if Uber were to reduce the quality of the Autocab service or force them to share their data”.

Read that again, because the regulator has just described the only protection that holds: not a promise, but the ability to leave.

Operators meanwhile asked the obvious question that week: what happens to our data? Autocab’s answer was that it already handles data for competing companies in some cities without sharing it, and that Uber would be no different. No abuse has come to light since, and that is not the point. The point is that the answer was a statement rather than a clause, while the real guarantee had already been named by the CMA.

April 2025. Lyft buys FreeNow for 175 million euros. It is not buying cars: ninety per cent of the value booked on that platform in 2024 came from taxis, which is to say from other people’s vehicles and licences.

In both cases whoever was driving stayed at the wheel. What changed was who sat on the other side of the contract: in the first the supplier became the competitor, in the second the app changed hands.

App first, then the car

The common objection in this trade is a fair one: nobody is going to sell you these cars.

Tesla, in its second-quarter 2026 update filed with the SEC, calls the fleet “our owned and operated fleet”: Cybercab in production in Texas and, as of July, service live in seven US metros. By October, said out loud on the call, the count was six cities. That filing says nothing about selling to anybody else.

Waymo does not sell its vehicles, and it has an app of its own, Waymo One, which serves Phoenix, San Francisco and Los Angeles. In Austin and Atlanta it chose to sit on the Uber app alone, and to leave Uber to manage the fleet, cleaning and depot included, which Uber in turn hands to a partner. Note the contrast with Zagreb, where the local operator owns the depot and Uber owns none of it. Waymo was not obliged to do it either way. It picked.

On 21 October 2026, on the third-quarter earnings call, the executives said both halves of it, and the first half cuts against us. Lars Moravy, VP of engineering: “we’re trying to get Cybercabs to customers by the end of this year, 2026”, and, as Tesla grows, “we’ll mainly deliver fully designed autonomous vehicles to our customers”. The ones built so far, he added, all go to the internal Robotaxi fleet.

Then Musk showed which way the balance tips: the Cybercab is “primarily going to be something that is offered as like transport as a service”, and “the vast majority” will stay Tesla-owned. The third-quarter update, filed the same day, names Cybercab for production and for deployment into the Austin fleet.

So yes, Tesla will start selling them, and saying otherwise would be dishonest. But on its own chief executive’s account the bulk stays theirs, and what they sell is the rides. Whoever sells the rides keeps the customer.

That order is worth reading twice. Nobody has ever had to rent a car in order to launch an app. Waymo, which has an app of its own, chose somebody else’s in two cities in order to reach customers.

In Europe a third model is taking hold in the meantime, and it is the one that touches an operator most directly: Pony.ai, Wayve and Momenta sell the autonomy to whoever operates locally, and in Zagreb the operator is a local company. It is worth looking at where those rides reach the passenger from. Pony.ai and Uber announced more than two thousand robotaxis for Europe in August, and in October announced London testing that had not begun by the time this piece was written; Uber has invested in Wayve, which works with Uber in London; and in Zagreb it has invested in the operator. The third model is real, but so far it reaches the passenger through the aggregator’s app.

The trade survives. The customer might not

In the European model somebody has to buy those cars, insure them, keep them on the road and make them pay. Not the manufacturer: a local operator, doing exactly what a private hire firm does today with a fleet of people. Knowing how to keep cars working has a future, and in Zagreb it has one with the aggregator on the share register too.

What does not survive, in any of the three models, is the customer nobody wrote down. And that is where this stops being about 2030 and starts being about next week.

Monday morning, in the office

The phone rings, somebody answers, the job goes out, the driver sets off. By the end of the day the work has been done and paid for. But of those passengers, what is left? A number in the call log, a name remembered by whoever was on shift, maybe a WhatsApp message on a driver’s phone.

A firm doing three hundred phone jobs a day has three hundred conversations. How many of those people could it ring tomorrow?

The regular who quietly stops calling makes no noise. Nobody notices, and by the time somebody does there is not even a list to work from. The same goes for the driver: the customer who rings his mobile directly is a valuable relationship, but it leaves no trace in the business and often none in his own next job either.

That is the real difference between you and an aggregator. Not the cars, not the technology. The memory.

From here on this is about what we sell, so read it as that.

What changes with UrbanGo, from day one

UrbanGo exists to close that gap, without asking an operator to change trade.

Every job that goes through the office lands in a customer record, the phone ones included, with a name, a number and a history. The caller who rings back in six months is recognised, and the one who has stopped calling shows up while there is still time to win them back.

The passenger who would rather book on a phone uses UrbanGo Passenger and ends up in your records, not a platform’s. Same convenience an aggregator offers, with one decisive difference: the contact stays yours.

Hotels, restaurants and local businesses book straight from their own dashboard, giving the name, the time and who to collect. The work arrives at the office without passing through a platform.

Drivers get jobs in an app, with the detail on them, instead of chasing them across radio, calls and WhatsApp groups. The fare is the driver’s and stays the driver’s. On every job it dispatches, the phone ones included, the office may take its own service commission: the driver who takes the job pays it, and sees the amount, alongside the cost of the job, before accepting. The coordination you already do every day stops being only a cost to you, and stops being an unnamed deduction to them.

And above all, customers, agreements, rates and job history export at any moment, in a format any other system can read, at no charge and without asking permission. The right survives you leaving, because that is what the contract says. It is not a reassurance given in a phone call: it is a clause, and a button in the software.

Why the question is not “who do I trust”

UrbanGo does not run transport, does not open its own offices, has no passengers of its own to protect and no interest in taking yours. Licences, shifts and the daily responsibilities are your trade. Ours is giving you the tools to keep it standing.

The Autocab case teaches one simple thing. When you choose the software that holds your customers, it does not matter who promises not to touch them. It matters who would lose something on the day they did. And it matters that you can take them away in one click.

In Zagreb the operator had the cars, the autonomy under contract and an app of its own, and still chose to sit inside an aggregator’s network. Not because something was missing: because that network brings passengers, and bringing them is the hardest part of the job. Here, for now, the customers still ring you. Having them is not the same as holding them, and the time is now, while it is still a choice and not a scramble.