Beeline held a Demo Day for journalists. What was discussed at the event

On June 30, Almaty hosted a closed Demo Day event for journalists and bloggers organized by Beeline Kazakhstan.

Beeline Kazakhstan is the country’s leader in subscriber numbers, with over 10 million Kazakhstani residents using its services. In recent years, the company has implemented a digital operator strategy and developed numerous digital products. The company’s portfolio includes the country’s first neobank Simply, music streaming service hitter, mobile television BeeTV, as well as business solutions for large industrial enterprises, ranging from IoT (Internet of Things) to Private LTE (private cellular networks). At the same time, Beeline is moving away from vendor solutions and expanding its own IT expertise. Currently, about 700 employees work in the IT direction, developing both external and internal Beeline Kazakhstan products.

The company periodically gathers media representatives to introduce them more closely to its ecosystem of digital products. This time, there were five presentations where employees explained why games and rewards appeared in the “My Beeline” app, why Beeline launched a music app, what results the Simply neobank team achieved in two years, how data scientists created a Kazakh language model, and what Big Data solutions the company is implementing with partners across Kazakhstan.

Everything, everywhere, all at once in the “My Beeline” app 

The first presentation was dedicated to the operator’s main app, “My Beeline,” which can rightfully be called a super app. 

10 years ago, the app was purely informational: you could check your balance, payment date, and call detail records. Afterward, we developed it as a service: for users, it turned into a service office that is always at hand. About two years ago, we shifted our focus to developing the app as a super app, so users could perform various operations in one place. Now, you can play games, make purchases, and get loans in the app. The next step — is personalized service based on the user’s experience with the app. 

The active user base of the app is 6.2 million, of which more than 4.1 million Kazakhstani residents use “My Beeline” monthly. Moreover, more than half of them log into the app to perform actions unrelated to their mobile number. 

Artem Loskutov, Product Manager of the app

In recent years, the product has seen an increase in new features. We created a flexible mechanism that allows adding new services and managing them through settings. For example, last year, a “Rewards” section appeared in the app, where users can perform various actions and receive rewards from Beeline in the form of free megabytes and gigabytes. For instance, if a user downloads the app and simply opens the “Rewards” section, they receive 200 MB of internet, and for opening a premium digital Simply card — 1 GB.

Additionally, games created by the Beeline development team are very popular among users. Just in June of this year, users played them 18 million times. According to the speaker, users are interested in games with national flavor, and there will be more of them in the app. Recently, tournaments also appeared in the games section: users play, earn points, and the best players get into the leaderboard and receive prizes in the form of free gigabytes.

Another innovation announced by the speaker is the ability to apply for loans in the app. This feature appeared at the end of last year, and since then, users have taken out about 4,000 loans. Currently, “My Beeline” offers loans from one bank, but soon users will be able to choose a bank with the most favorable terms for their needs.

Artem Loskutov, Product Manager of the app

Throughout the entire time working on the app, we have come to better understand the profile of our user. Previously, we thought that offering everything at once was the formula for success that would meet all users’ needs. But in the rush to do everything, we focused on the quantity of what we release rather than on users’ needs. Now, we daily research their needs and direct our efforts toward what they truly need.

Two years for neobank Simply: what it lives by, what it creates, and what we expect

The neobank Simply, the first in Kazakhstan to offer an innovative solution—a symbiosis of telecommunications and banking services on a single platform, celebrated its second birthday. According to Marat Zhagiparov, the product owner, more than 1.6 million Kazakhstani residents have downloaded the neobank’s app during this time. Purchases with the card were made in 93 countries on all continents of our planet, with the most distant one being in Chile.

With the launch of the financial service, the company has continuously expanded user capabilities: constant cashback from 1.5% to 10%, matching the card number with the phone number, integration with Apple Pay, Google Pay, Samsung Pay, Garmin Pay, and the ability to send international transfers of up to nearly one million tenge without fees. Customers can also use the privileges of the premium digital Visa Platinum card for free. For example, they can visit Lounge Key business lounges in airports and pack their luggage in Almaty and Astana for free. Simply customers have already used the lounge service 4,169 times and luggage packing 1,041 times.

Over two years of operation, the total amount of cashbacks issued by Simply has approached two billion tenge. According to the neobank, the most popular payment categories for Kazakhstani residents were supermarkets, gas stations, electronics, public catering, restaurants, and fast food. 

Another innovation of the neobank is sending split bills via the Request to Pay service. Now, customers can split a bill by sending it to multiple recipients at once, specifying the amount for each. With this feature, Simply neobank became the first fintech company in the world to launch this solution based on the Visa Direct platform.

Thanks to the work done, the service has received numerous awards. At the end of 2021, in its first year of launch, the company received three awards from Visa: “First Fully Digital Neobank Card,” “Best Premium Issuer,” and “Google Pay Launch.” And in 2022, Simply received a Visa award in the “Best Fintech” category.

Speaking about plans, Marat noted that Simply users will be able to choose individual card designs in the app and Apple and Google Pay wallets, manage subscription services, and receive bonuses for using Beeline ecosystem products.

Big data guarding Kazakhstan’s enterprises: how computers learned to count chickens and guard people 

Khakimzhan Kuvatov, Head of the BigData Business and Open Ecosystem Development Department, talked about business solutions developed by Beeline’s Big Data team.

Computer vision is a technology that allows analyzing and extracting information from video. Video analytics, in simpler terms, is an attempt by developers to train AI models to recognize almost everything in a video stream that the human eye sees and recognizes.

Using examples from several industries, Khakimzhan explained how video analytics simplifies people’s lives and improves business operations.

Retail sales. Using video analytics models, it is possible to analyze customer flow: study visitor numbers, demographic indicators, analyze emotions, and also conduct employee-customer interaction analysis to improve business metrics. At the same time, system users do not study the video themselves but receive information in the form of statistical data and infographics, which can be analyzed to improve processes. 

Industry. In industries of this category, video analytics ensures industrial safety. First of all, the system allows monitoring whether employees follow safety rules, use personal protective equipment (PPE), and so on. Also, in case of people falling for various reasons at industrial facilities, the system notifies the responsible employee so they can learn about the incident in time and help their colleague.

Another example provided by the speaker is monitoring the condition of bucket teeth when working with rock. For instance, if a tooth enters a rock crusher, the latter may break, leading to costs for replacing expensive equipment and downtime at the enterprise.

Agriculture. Khakimzhan gave an example of counting chickens and analyzing their behavior. Thus, the model can identify birds and count their total number.

You can learn more about video analytics here.

All the models mentioned by the speaker are developed independently by the Beeline team to meet customers’ needs. The company handles this issue with a local team, which is a fairly large staff of specialists. The big data team consists of about 100 people who have already implemented a large number of models based on the big data available to Beeline. This is the main advantage as a provider of such services.

Connecting art and technology: why we launched the music app hitter

In the summer of 2022, Beeline Kazakhstan launched the music streaming service hitter for all Kazakhstani residents, whose mission is to develop and support the Kazakhstani music scene.

Nigora Bakhovudinova, Product Marketing Manager

We focus on local music content, continuously working on improvements to promote Kazakhstani music. We create curated playlists, give emerging artists the opportunity to be heard, and develop our own community. At the same time, we understand that we are in a global music environment, so our library contains millions of tracks from the international scene.

The fact that the right product direction was chosen is evidenced by the numbers: 80% of all tracks listened to in 2022 were by Kazakhstani artists. The top 3 most listened-to tracks in the app: Qarakesek «Қураған гүл», Moldanazar Janym Sol, Musicola «Певица и саксофон». Currently, there are 260,000 registered users in the app.

In a short period of existence, the team has implemented several significant projects. One of them was a competition for emerging artists together with the Gakku Dauysy festival, which took place last year. The winners of the competition, Sannur & Collid-e, performed on the main stage of the festival in front of hundreds of thousands of viewers with the track UUU. This year, hitter supported the album hotel amour by artist abdr., which was exclusively presented in the app a week before the official album release.

In 2023, the team plans to improve the app for users as well as grow useful functionality for musicians and people from the creative community. 

How and why we created and released the Kazakh language model to the public

Elmira Badretdinova, Product Owner of conversation platform (or bots) Beeline Kazakhstan

Beeline Kazakhstan uses various channels for interacting with subscribers: call centers, chat bots in our apps and on the website, as well as messengers and social networks where users can submit requests. Since around 2015, the company has had a chat bot for text requests, which successfully handles about 200,000 requests per month. 

However, most user requests come in Russian. But in recent years, according to Beeline data, the number of requests in Kazakh has been growing. Since 2021 alone, this figure has more than doubled.

Taking into account the bilingualism of most users and the common habit of writing requests in Kazakh using a Russian keyboard layout, using, for example, “k” and “g” instead of “q” and “gh,” there was a need to improve the chat bot in Kazakh. Thus, in April of this year, the company’s development team created and released the Kazakh language model BeeBERT to the public. This was discussed by Big Data analyst Beksultan Sagyndyk.

Kaz-RoBERTA-conversational (BeeBERT) is a language model trained on a large dataset in Kazakh. BeeBERT is applicable to text information and predicts which word was missing in the text. 

The published language model is not a final product that users can use independently, but now any developers of language models can integrate it into chat bots and content analysis systems. With further refinement (fine-tuning), BeeBERT will be able to translate, summarize, simplify, and understand the tone of texts in Kazakh.

BeeBERT has improved the quality of text recognition in Kazakh in the company’s chat bot and enhanced the ability to automatically determine the language of the chat bot interface based on each customer request. For example, if one request was in Russian and another in Kazakh, then, thanks to the model, the chat bot interface language will change according to the language of the request.

Since its publication on April 6, the model has been downloaded more than a thousand times. Publishing such a model to the public is the first case among major technology companies in Kazakhstan. This step should serve as a good example for other companies. By publishing the model to the public, the company wants to show its openness and contribute to the development of the language direction of artificial intelligence.

The material was prepared with the support of Beeline Kazakhstan.