Ten years ago the dispatcher was the center of a taxi service. They took the call, wrote down the address, picked a driver over the radio and made sure the order didn’t get «stuck». Today each of these steps can be handed to software. The service keeps running around the clock, drivers get orders, customers see the price and pickup time, and a person steps in only where they are truly needed.

In this article we look at what «taxi without a dispatcher» means in practice: which processes get automated, how order dispatch works, what stays with people, and what risks a service faces when it switches to this mode.

What a dispatcher actually does

To see what can be automated, it helps to break the dispatcher’s job into separate operations. In a typical service in a small or mid-sized city, these are:

1. Taking the order. Answer the call, confirm the pickup and drop-off addresses and any requests (child seat, luggage, cashless payment), record the phone number.
2. Calculating the price. Quote a price, or at least a rough figure based on the rate.
3. Finding a driver. Work out which driver is closest and free, offer the order, wait for confirmation.
4. Informing the customer. Give the car’s plate number and color and the pickup time, then later say the car has arrived.
5. Monitoring the trip. Make sure the driver set off, arrived and completed the order; if something goes wrong, reassign it.
6. Handling exceptions. The customer doesn’t come out, the driver cancels, the address can’t be found, a dispute over the price.

Items 1–5 are routine that repeats hundreds of times a day following the same script. This is what gets fully automated. Item 6 covers one-off situations that require judgment. They can be automated only partly, and that’s fine.

How order intake works without an operator

Full automation starts with channels where customers can place an order themselves, without talking to a person. UpTaxi has several such channels, and all of them are included in the base package:

• Customer mobile app (Android and iOS). The customer enters addresses, sees the price upfront, can pay by card, place an advance booking and view their trip history.
• Dispatch service website. A ready-made order form for those who don’t want to install an app.
• Order bot in Telegram and Max. Ordering in a familiar messenger, with nothing to install: the customer sends their location or types an address with suggestions (train stations, shopping malls), chooses a class and options — economy, comfort, child seat — and confirms. Then the chat receives statuses «driver found», «car has arrived», «on the way» with the driver’s car make, plate number and phone.
• LLM voice robot operator. For those who call. The robot answers the call, asks where from and where to, clarifies details, understands natural speech and pauses, and creates the order without an operator, around the clock. This matters for the audience that still calls rather than taps buttons — and in small cities they make up most of the customers.

The key point: all channels write orders into one system. There’s no situation where app orders sit in one place and phone orders sit in the operator’s notebook. More on intake channels on the Receiving an order page.

What happens when a customer calls anyway

Even in the most automated service, some customers will call. This used to mean you couldn’t do without an operator: customers wouldn’t put up with robots with a rigid «press 1» menu. A robot built on a large language model works differently — it holds a conversation, not a menu. The customer says «from 15 Lenina Street to the train station, in about ten minutes», the robot asks for the entrance, confirms the price and creates the order. If the customer goes quiet, interrupts or changes their mind, the robot handles it instead of dropping the call.

This is the key difference from most software on the market: only a handful of competitors have voice robots, and as far as we know, none of them is built on an LLM. For a service, the difference is measured by the share of calls the robot turns into orders without transferring to a person.

Options that don’t require your own operator staff:

Option	How it works	Who it suits
LLM voice robot	Answers the call, understands natural speech, clarifies details and creates the order with no human, 24/7	Any service whose customers call; covers the main flow
Distributed call center	The call is taken by an operator from another service within the system; costs 1 ₽ per accepted order and 4 ₽ per completed order	Small services — for calls the robot passed to a human
Night call forwarding	Your own operators answer during the day; at night, the robot or other services’ operators	Services cutting their night shift

The distributed call center in UpTaxi is built as eight independent, competing call centers within one system. A small service buys call handling; a large one earns money from its operators by taking other services’ calls. Details on the Multi-console and call center page.

Order dispatch: why «straight-line» is bad

Once an order is accepted, the system has to choose a driver. The simplest way is to take the coordinates of the customer and all free cars and calculate the straight-line distance. Many programs work this way, and this is exactly how the situations every dispatcher knows arise: a car «300 meters away» takes 12 minutes to arrive because there’s a river, a railway or a one-way street between it and the customer.

UpTaxi calculates distance differently — over the city’s road graph. This means:

• the real road network is taken into account, not geometric distance;
• one-way traffic is taken into account;
• prohibited turns are taken into account;
• it uses its own routing algorithm, without calling external Yandex or Google services.

As a result, the closest car is the one that will actually get there fastest. For a service, this means two measurable things: shorter pickup times and less empty mileage. The second directly affects the driver’s earnings — and therefore whether they stay with your service. Technical description on the Automatic dispatch and route algorithm page.

Two ways to deliver orders to drivers

In the «UpTaxi Driver» app, orders can arrive in two ways, configured at the service level:

1. Order list. The driver sees available orders and chooses. This mode is familiar to drivers who came from ride-hailing aggregators and suits services where drivers use their own cars and value freedom.
2. Auto-assignment. The system assigns the order to the nearest suitable driver. This mode gives the fastest pickup and removes people from the chain as much as possible.

In practice, many services combine them: auto-assignment for the main flow and the list for «complex» orders — intercity, advance bookings, orders with special conditions.

Drivers without the app: the bot in Max

A separate case is a driver who doesn’t want to or can’t install the app: an old phone, little storage, weak internet outside the city. For them, UpTaxi has a driver bot in Max — the national messenger that has come preinstalled on smartphones sold in Russia since 2025. The driver opens the bot and gets the same set of features: automatic dispatch, a feed of open orders, statuses, editing addresses in an order, chat with the dispatcher. The bot works on weak internet and uses minimal data; if GPS isn’t available, the driver marks their location manually. Bot example — https://max.ru/id890603474200_2_bot.

For automatic mode, this means the barrier to onboarding a driver becomes minimal: nothing to download, update or configure — the messenger already on the phone is enough.

Want to see how it works on your orders? Services with 50+ completed orders per day can get a free 30-day trial — no license purchase and no server. Terms on the pricing page.

What happens if nobody takes the order

The most vulnerable point of automation is an order nobody accepted. A dispatcher in this situation has a phone and persuasion. A system needs rules.

UpTaxi handles this on several levels:

• Re-offer. If the first driver declined or didn’t respond, the order goes to the next one by distance.
• Surge pricing. If there are more orders than free cars, the price goes up automatically (or manually by the manager), and the order becomes more attractive to drivers.
• Driver exchange. If your own drivers don’t take the order within a set time, it goes to the exchange, where a driver from another service can complete it. The customer and the order data stay with your company — the «outside» driver only does the trip.

This builds a chain with no manual intervention: own drivers → higher price → other services’ drivers. The service doesn’t lose the customer even if it doesn’t have enough cars of its own at that moment.

Informing the customer without a human

The second-largest block of routine is telling the customer the order is accepted, a car is assigned, it has arrived, and the trip is complete. UpTaxi uses voice and SMS auto-notifications based on the Asterisk PBX for this: the system calls or sends a message according to a set script. In the app, the customer also sees the status and the car on the map, and in the Telegram or Max bot, statuses arrive right in the chat — with the driver’s car make, plate number and phone. Sending promotions and promo codes to customers through the bot is free, unlike SMS.

For the service, this means operators don’t need to call back to say «the car is at your door». For the customer, it means they don’t have to wait for a call and guess whether the taxi has arrived.

What stays with people

«Without a dispatcher» doesn’t mean «without people». The nature of the work changes: instead of a flow of routine operations, what remains is handling exceptions and management. It makes sense to leave these to a person:

• Disputes. The customer says the car never came; the driver says they waited 15 minutes. You need a person with access to the order history and the GPS track.
• Non-standard orders. Transporting pets, delivering documents, intercity trips at a negotiated price.
• Working with drivers. Onboarding, briefing, reviewing complaints, deciding on blocks.
• Managing rates and economics. When to turn on the surge multiplier, what commission to keep, how to set up bonuses.

All of this is done from the manager’s account, which brings together statistics, reports, and management of rates, drivers and payroll. So one person with a laptop covers functions that used to need a shift of several operators.

The economics of switching

Let’s take an example. A service with 300 orders per day keeps three operators in shifts to cover 24 hours. At a salary of 40,000–50,000 ₽ (typical for a small city) per month per person, that’s a payroll that doesn’t depend on order volume: in a bad month the operators cost as much as in a good one.

When switching to automatic mode, these costs are replaced by:

• the software’s daily fee and commission (in UpTaxi, for 101–300 orders per day this is 150 ₽ per day and 9% of driver commission; from 501 orders there is no daily fee);
• the cost of handling «live» calls through the distributed call center — only for orders that actually came in by phone;
• optionally, one employee to handle exceptions on the day shift.

The main difference is that costs become variable and grow only with orders. In the off-season or when demand drops, the service doesn’t pay for idle operators.

Common mistakes when switching

From the experience of services that moved from radios and old software to automatic mode, three things get in the way most often.

Trying to do everything in one day. Drivers used to the dispatcher’s voice see auto-assignment as «a robot taking away orders». It’s smarter to start with the order list, then turn on auto-assignment during peak hours, then go fully automatic. Sevtaxi from Sevastopol, by its own estimate, got its dispatchers adapted in about 10 days and was fully «on track» after 1–1.5 months — a realistic timeline.

No plan for exceptions. If nobody has defined who responds to a price dispute or a car that didn’t show up, and how, automation turns into «nobody answers». Assign someone responsible and give them access to the order history.

Wrongly set intervals. How many seconds to wait for a driver’s response, after how many minutes to send the order to the exchange, at what car shortage to turn on the higher rate — these parameters need to fit your city. Start with conservative values and adjust based on reports.

What you need to launch

Full automation doesn’t require a server room or IT staff. To launch UpTaxi you need:

• internet from 1 Mbps;
• a computer with 4 GB of RAM and Windows 7 64-bit or newer;
• a headset for the operator (if you’ll still have one);
• a SIP trunk (or GSM modem / VoIP gateway) and a multi-line number;
• for drivers — Android smartphones with GPS for the app, or the Max messenger for the driver bot.

No separate server or mini-PBX is needed: the server for your service is set up in the cloud within a day, installation and configuration are done remotely, and technical support works 24/7 and is included in the price. The system has been tested at a load of 25,000 orders per day for a single taxi service and 900 drivers online at once, so there’s room to grow.

In brief

• Dispatch routine — intake, pricing, assignment, notifications, monitoring — is fully automated.
• Customers order the way they like: a call (the LLM robot answers), the app, the website, a bot in Telegram or Max. Drivers work from the app or directly from Max with nothing to install.
• The quality of automation depends on the dispatch algorithm: road-graph calculation gives real pickup times, «straight-line» calculation gives an illusion of proximity.
• For «unfilled» orders you need a chain: re-offer → surge pricing → exchange.
• People are left with exceptions and management, not a flow of calls.
• It’s better to switch in stages: order list → auto-assignment at peak → full automatic mode.

If you want to test automatic mode on your own orders and drivers, request 30-day access. For services with 50+ completed orders per day the trial is free; terms and rates are on the pricing page. You can ask questions on Telegram, on Max or by phone at +7 (900) 28-28-328.