Ural Airlines
Ural Airlines

Ural Airlines: A bot to process up to 63% of all requests

Ural Airlines used a call center for all incoming customer requests. Flomni made it possible for the company to optimize performance and improve customer service quality


Every month, up to 63% of all customer calls are processed automatically, with no human operators involved
The Smart Bot boosts the chatbot by 28% thanks to a random text recognition module
Call service costs have been reduced by 25%


Ural Airlines has been engaged in domestic and international passenger transportation since 1993. The company's services are used by 9 million passengers. It is one of the top largest airlines in Russia.


Passenger support used to be provided by the company hotline only. Call service was expensive, and users were not happy with it: many people prefer text messages to phone calls, and for some, calling is simply inconvenient, for example, if the passenger is in roaming mode.


Implement and automate text-based message channels to process requests.


Omnichannel platform with Smart Bots

In order to build a unified system for communicating with customers, we introduced several Flomni products.

The project was implemented on a turnkey basis with minimal involvement of the company.

  1. We connected an online chat widget to the website and integrated the chat feature into Ural Airlines mobile app through SDK.
  2. The website widget included both an online chat and links to messenger apps that passengers could use to contact the company. We selected the most popular messengers — WhatsApp, Telegram, and VKontakte.
  3. We connected chatbots and Smart Bots to all chats. This means that the bot will respond first no matter how customers send a message — via an online chat or a messenger app. This processing format helps automate most responses.
  4. We connected the Dialogs platform to receive requests from all communication channels. Operators now work in a single UI and can respond to all messages handed off by the chatbot in the order of arrival.

We quickly deployed the chatbot on the company website and in the mobile app. I really like the attitude of everyone who works on our project. We get prompt, qualified answers from professionals to all our requests. Rather than provide dry formal answers, they give real advice that actually works.

Yulia Shibina

Head of Marketing at Ural Airlines

How we developed the bots

To implement bots into any project, you should take a number of crucial steps. Let's use Ural Airlines as an example.

  1. We identified the most frequent topics of customer inquiries. These should be the first to be automated by bots.
    Flomni specialists worked on it together with a customer service employee at Ural Airlines. There were 14 topics identified in total: flight status, ticket refunds, ticket exchange, etc.
  2. We studied request processing scenarios.
    A bot should process requests similar to how a contact center operator would do it. To do this, we looked at the scenarios used by contact center operators.
  3. We implemented said scenarios into the chatbot logic.Flomni specialists built the Q&A logic
    and developed text responses for chatbots based on the resulting terms of reference. Example: passengers often want to make changes to a ticket they have purchased. A human operator would have asked for more details about the ticket and provided all possible solutions to the issue. All the same questions are asked by a bot, which then responds to the passenger's chosen answers.
  4. imageMeta.blog/content-ural-1
    This is how the button bot works once the Q&A logic is clear.
  5. In case of button bots, that would be enough. But if a company, like Ural Airlines, wants its passengers to be able to send a question right away instead of choosing from several options, they might need a neural network bot. A bot like that needs a training database.
  6. We trained a neural network to recognize random text.
    For each of the chosen topics, it had to select from over 50 sample requests. From them, we identified different versions of wording the same question, so that the bot could understand them all. Based on the results of the pilot launch, it became clear that the solution worked well in terms of user experience. Neural network bots processed requests correctly, thus reducing the call center operator workload.

Work of contact center operators made easy

Now all text-based requests are processed by Dialogs, a unified chat platform. This is where customer requests from all channels flow in. At level one, all requests are processed automatically. At the same time, Ural Airlines specialists can, if they wish so, look through bot-processed requests.

If a bot fails to handle a request, or if a passenger wants to communicate with a human employee, the chat is handed off. Operators can take a look at the entire chat history between the customer and the bot. This way, they don't need to ask for details — they can go straight to the point and answer the passenger's question. Operators use Dialogs to respond, and their responses are then forwarded to the customer's channel — for example, VKontakte or a messenger app.

This is how customer requests are displayed in Dialogs.

If this is not the first time this customer contacted the company, the operator will know this. The request history will be displayed in the chat window. Moreover, if a customer first uses VKontakte and later Telegram, both requests will be identified as requests of the same person, tied to a single chat, and served as one. These two questions coming from two different channels and being processed by two different contact center employees is no longer a possibility.

The customer still has the option of choosing to chat with a human operator; in this case, the chat will be handed off to a company employee after a few clarifying questions.

Passengers can now choose a convenient communication channel, ask a question in a free form, and switch to a chat with a human operator

Extended bot capacity

It goes two ways.

  • The neural network bot improves text recognition on topics that have already been implemented. The Flomni account manager assigned on the project regularly analyzes the list of requests that the AI bot has failed to process. Then, they identify the relevance of each request to a particular topic, and add the new wording to the system. The bot can manage on its own now that it has learned the wording.
  • New themes are being added to the bot's service, as requested by Ural Airlines. When an AI bot encounters requests like these, it determines the topic without assistance and processes it according to a preset algorithm.

Extended bot capacity

Ural Airlines have now found that the chatbot is efficient and intend to expand its capacity in the future.

  1. Connect outgoing notification processing.
    After that, customers can be notified about the status of their flight in messenger apps. With these notifications, customers will be able to manage the booking. For example, they can order extra meals or choose a seat.
  2. Integrate Flomni with the CRM system at Ural Airlines.
    This will help the company recognize which flight the customer has booked and what their question is related to by their cell phone number. This can speed up the service and boost service quality. Similar CRM integrations are already proving to be highly efficient in other Flomni projects.

Gradual expansion of bot capacity, as in this case, allows businesses to quickly automate key processes. Any decisions related to further project development should not be based on forecasts but real-time results.

Get free advice

Our experts will answer all your questions
and help you find the right solution

decor left
decor left
By pressing Submit, you agree to the Personal Data Processing Policy
decor left
decor left