A chatbot may not give the desired effect to your business. Which companies definitely don’t need it? I made a bot but got confused

First, we identified the main office processes. We didn't even talk about the chatbot. For example, ordering business trips. Now I have to write to the business travel service employee in the mail: “I’m going on a business trip to Surgut on the 5th for three days,” and he: “Such and such a plane and such and such a hotel - is everything suitable?”, and I: “ Come on". Then he will go to coordinate with the management, book the tickets himself, and after some time he will send me confirmation that everything is OK. A bot can do the same thing.

Or if you need a certificate for a visa, then the bot will knock on the bus, the bus will knock on the personnel subsystem and pick up the PDF, then send it to the HR department printer and write a letter that a stamp is needed there. Then he will notify me that I can come back in a couple of hours. If you need to issue a pass for a guest or book a meeting room for a meeting, then you just need to entrust this task to the bot, and it will complete it.

Now let's show an example a little more complicated.

In 3 weeks:

Our dialogue about a business trip uses a standard business process. The employee (his full name is in the authentication data) indicates the city of the business trip and the dates. The bot writes a ticket to the accounting department and a letter to the specialist responsible for purchasing tickets and making travel reservations. When the ticket is closed, the bot receives its result and further processes it according to the script. The bot processes the specialist with letters or Telegram messages until he issues documents with tickets and reservations. It then searches the user's office Wi-Fi hotspot to determine the nearest available printer. Sends documents for printing. On the day of departure, it reminds you about the trip, takes the weather from one of the external sites according to its open API, then begins to query the airport’s online board for the flight schedule (the flight itself is on the tickets). After the trip, he will ask for a report on funds for accounting (he will send a form and tell you how to fill it out), plus he will ask useful tips for the knowledge base. In addition, he can order a taxi to the airport, order a car on the spot and collect trip data to enter the amounts directly into the expense report.

As you can see, each step is quite simple, but together they create a bit of magic. The only downside is a lot of fiddling around with documents, various exceptions, and logging into systems that are not yet ready for such a digital transformation. A lot of functionality is still in beta, alpha, or even just at the integration stage - when working with documents, development and testing take a very long time, so that an accidental woodpecker does not bring down the entire office. But everything looks quite simple and logical.

A little more magic

Colloquial speech
Bots can learn from user syntax. Here is a simple example of what a call to Help Desk looks like through a bot:

And this is a screenshot of an application in ITSM:

The process now is like this: the user sends a more or less structured ticket, and then the help desk employee breaks it down into type and subtype, after which a deadline for the incident and a person in charge are assigned. Next, you need to indicate the type of request - incident. Service - printing devices. Employee - Voronova. Responsible - Petrov. Office - 1105.

Over the course of several years, quite a few pairs of free-form verbal requests and correctly filled out ITSM system forms have accumulated. Then everything is simple: a little machine learning on this sample, and then the bot, together with the user’s ticket, sends the type and subtype of the application to the system. Naturally, he may make mistakes, so the process has changed slightly so far - the help desk employee simply receives the fields that have already been filled in (in 80% of cases they are filled out correctly), but, of course, he can change their values.

An administrator can also use a bot in his work, for example, for this:

Voice

IVR systems are the same chatbots with a slightly modified interface. When you connect synthesis and voice recognition systems (like these, there are also examples of dialogues), you can get the same chat in voice form. Many users really like this, and now some office automation functions work through IVR. This will be especially relevant for those customers where users are not very familiar with computers.

Integration with any processes for many users

We showed the bot at the exhibition, and the idea of ​​tight integration with the office bus has already spread to customers. For example, one of them wants to launch telegram messages to coordinate a server reboot. The idea is this: the request is approved by a number of people, when a lot of “ok” responses are collected, the server is rebooted. In general, this is quite realistic. Posting a document can be done in the same way. But this is already a very distant future.

For example, a chatbot, human or IVR is a data interface, the user task is “order a meeting at 15:00”. The connector is an authorization link. The bus carries out the implementation of the task.

More opportunities

Another completely banal example of how the bot works is account recovery. Users (and there are about two thousand of them in the office) regularly forget their passwords. Either they activate accounts on several devices - a phone, a tablet, a workstation... And then either they forgot to update the password somewhere, or the tablet was left at home and the child locked the account. In general, the moment of changing the password is a headache both for the employees themselves and for the technical support, which they often call (especially in the morning). Now the bot helps you reset your password, or rather, unlock your account. The operation is absolutely routine.

Address book search:

Here is a complex analysis of the query phrase for searching for a contact within the company. Next is a request to the HR service, where we carefully rewrite all the useful skills of people in case of new projects and urgent requests. “Picani” is a fairly rare word, so searching for it is very relevant and clarification is not required. The bot uploads the data of an employee who understands the issue. If he were online, you could immediately write to him. He is offline, and judging by the documents of the same personnel department, he is on a business trip in Surgut. The bot checks its time zone and provides data.

Everything here is also relatively simple. You can tell the bot “remind me about lunch with Ivanov,” and it will set this as a task to remind you in time. The remaining data is available on the bus.

How it all began

It all started with the idea of ​​finding out what digital transformation, digitalization, etc. is, otherwise there is a lot of noise in the market, but no specifics. In order to understand what it is, we decided to start with ourselves and understand what can be “digitalized” in our company. While “digging the topic,” we worked through a bunch of cases for digitalization, found solutions, but at the same time we were haunted by the idea of ​​the “correct” interface for office business processes. We didn’t even talk about the chatbot, in principle it wasn’t even discussed. At some point, we realized that the approach to interaction with office business processes also needed to be changed, everything should be convenient and necessarily mobile, and since 100 mobile applications cannot be made, it became clear that the future belongs to natural interfaces, in particular chats. In particular, there is the incredible Chinese example of WeChat - originally an analogue of ICQ, which has now become both a social network and trading platform, and notice board, and by email, and in general to everyone.

Bots can also tell basic things from the corporate training knowledge base, search for training materials, show the nearest internal courses and seminars, play the list of achievements with newcomers (“Got my stationery”, “Found the kitchen”, “Checked in at the gym” and so on) , book a rest room or stand in line for game console in the same place, get a password for guest Wi-Fi, find out the number of the VHI, clinic number, ambulance phone number, inform the office medical service about an injury, make an appointment with the office medical service, book a simulator, report a lost or found item (the bot then connects people or drops the log secretary), send corporate news, give various information new employees, how to live, warn about the birthdays of your contacts. Long-term plans include checking the length of the queue in the canteen using video analytics (usually from 1 to 15 minutes, now this is implemented by a separate informer in our office), finding out the menu for today in the canteen or ordering pizza to the office, finding travel companions, handing over documents. From technical service The plans are to replace the corporate SIM card, change the tariff, accept bug reports for any things like “here on the fifth floor the lamp does not light”, send a bug report to the housekeeping service or the cleaning staff.

Of course, the main stumbling block for all customers with this approach to office digital transformation is the security (authorization) of users, and the second is the place where all this correspondence is stored. In the case of our bot, correspondence is stored in the CROC cloud, on the bot server, that is, in our internal infrastructure. All integration between the bot and the company’s systems occurs through a single integration bus. The connection to the bus is through a universal connector, which allows you to divide access rights into three parts: publicly accessible parts (public), for all employees and for specific employees.

Today there is no single “educational standard” for chatbots. There are bots that can perform only a few functions, for example, placing an order in an online store and finding the nearest pickup point. But there are also complex systems, using artificial intelligence and predictive analytics algorithms to predict what product and at what price their interlocutor is ready to buy.

So what can chatbots do in ecommerce and is it true that they will soon put salespeople out of work?

Selection of goods by parameters is perhaps the most common function of chatbots. The bots from retailers H&M and Sephora are the best at handling it today. The most interesting thing here is the algorithms with which bots gradually learn everything more information about a person and more accurately recommend products to him. Otherwise, chatbots can perform virtually any e-commerce operation if they are integrated with the company’s database.

Chatbotsare already capableduplicate a live consultant and help the user decide complex tasks with a branched election tree. For example, help you choose a laptop or tourist trip, when initially a person does not know exactly what he needs. perform such functions. In addition, the “correct” chatbot understands when it can’t cope with a task and it’s time to switch the user to a live consultant.

Chatbots vary in their level of proficiency in so-called "natural language"- that is, the language that you and I speak. Some chatbots do not speak the language at all and communicate with users, asking them to press buttons or choose from answer options. Others, using special technologies, analyze people’s remarks and understand their requests, even if the phrase is formulated with errors and in an informal, conversational style. Many chatbots are also able to generate lines so that their dialogue with a person sounds natural, and the same wording is not repeated. We also work with virtual bots that can understand not only text, but also speech spoken over the phone, and synthesize speech in response.

It is important to understand that there is no single level, which any developer or customer of chatbots should strive for. It all depends on the task that is set for the system in each specific case.

The “top” of the professional hierarchy of modern chatbots are systems built on machine learning, which “draw conclusions” from each dialogue and constantly improve their own efficiency.

What do chatbots do best?

• Sales in messengers. Messengers are now the favorite digital tool of many people, both in Russia and abroad. In our country, for example, up to 85% of Beeline clients use instant messengers, often several at once. Chatbots allow you to organize the entire sales process in the messenger - from product selection to payment and parcel tracking. In this case, a person does not need to open a website or install an application, study the store interface, and the whole process resembles a regular conversation.

• User support and communication24/7 on all channels. IN modern world An e-commerce user expects an almost instant response to any of his requests. According to , due to support issues and customer service companies lose up to three out of every five customers. But 24/7 support it costs expensive. Chatbots answer user questions and solve their problems 24 hours a day, and they cost several times less than an army of call center employees.

• Providing marketing funnels. When it comes to converting leads into converted users and repeat purchases, chatbots have many advantages over email marketing and other traditional channels. Bots can instantly evaluate the user during a dialogue and change the script. They “live” as “friends” of users in messengers and can unobtrusively inform them about special offers. Finally, chatbots do not have “abandoned carts”, the owners of which must then be caught using remarketing. If a person interrupts the dialogue, then later the chatbot can simply remind him of the unfinished purchase.

• Solving standard, repetitive tasks in any business processes. Modern chatbots cope with such tasks as checking the status of a customer’s order in the database, changing the size of an item already purchased in an online store, and checking issued invoices. It's not just about communicating with clients. Many chatbots successfully work in internal business processes: collecting information about hours worked by employees, creating work shift schedules, and so on. With the help of chatbots, businesses save hundreds of working hours and millions of rubles and allow employees to do more interesting things than standard procedures.

How to start a chatbot from scratch

Typically, you should start by creating a chatbot with a limited set of features(e.g. product selection, sales, special promotions and support), but with the potential for development. Don’t forget that chatbots bring real benefits when they are integrated with the company’s internal systems (CRM, product catalog) and external services(newsletters, analytics).

Before you start developing a chatbot, you need to decide what tasks you set for it and discuss its future functions in detail with employees of the relevant departments. Then you need to choose a developer. We recommend interviewing several developers, comparing prices, solutions, and paying attention to hidden costs. For example, many companies can develop a chatbot for you, but do not provide the tools with which you can manage it internally. Then for any change in the script (and you will definitely want to make such changes when you analyze the bot’s work) you will have to pay. Try to find a developer who will provide you with the most flexible tools for analyzing results and managing a ready-made chatbot.

Chat-bot on the site vs. chatbot integrated intopopularmessenger. Whatchoose?

It all depends on what communication channels your audience prefers and what functions the chatbot performs. On websites, chatbots act as consultants, answer frequently asked questions and help you find faster required pages. Chatbots in instant messengers can do their work without the help of a website. Analyze where your users “live” and how a chatbot can make their lives easier in the channels they love.

Our favorite case study on this topic is Thai Knorr, who realized that the main problem of housewives is figuring out what to cook for dinner, and what they discuss about it at the local messenger Live. Knorr came up with a bot that suggests suitable recipes, installed it in this messenger, and sales of bouillon cubes increased by 50%.

Where chatbots are ineffective

More precisely, which approaches to using chatbots are ineffective or even destructive.

You should not use this technology to simply “spam” users in instant messengers. Those who expect that a chatbot can 100% replace a human, for example, in support, will also be disappointed.

Automatic interlocutors can take on only up to 70% of requests, and the rest will have to be dealt with by real people. You need to be especially careful when developing a chatbot that gives people advice, for example, on choosing medications or solving psychological problems. You should always warn that a robot cannot replace a living person and do everything to avoid lawsuits. It is also necessary to pay attention to security - not only of payment data, but also of user personal data.

How"teachingut" chatbots

Chatbots that understand natural language use a complex, multi-level algorithm for analyzing what is said - at the level individual words, grammatical structures, even emotional messages. As a result, they are able to understand what exactly a person wants, clarify their understanding with the help of leading questions, and launch one or another scenario (for example, selecting a product).

The scripts themselves are ideally flexible– the chatbot understands the client’s characteristics right during the conversation and changes tactics, personalizes the conversation, and offers other products. Scenarios are developed in accordance with the wishes of the retailer, its business processes and marketing know-how. With each next step in the dialogue, the chatbot must take into account many factors: how it understood the client’s previous response, the entire context of the dialogue, the history of the client’s previous requests, the availability of goods in the warehouse, and so on.

At original development chatbot algorithms for understanding natural language are developed by linguists taking into account the specifics of a particular business, product group, and so on. In the first weeks of a chatbot’s life, a team of linguists monitors it and further trains it.

How many people visited the siteonline stores start talking to a chatbot? What is the conversion from conversations to orders?

This data varies from case to case and depends on the type of business, the functions of the chatbot, and how well its creators were able to guess the needs of customers. On average, chatbots have a very high percentage of conversion of leads into conversation partners. If the average open rate of e-commerce mailings is 25%, then in the case of chatbots, 80% of recipients respond to the message and enter into dialogue. This data is provided by the American platform for developers Pandorabots, which hosts thousands of bots for different verticals. If we talk about conversions from conversations to purchases, X5 Retail Group has achieved excellent numbers with a chatbot that selects New Year’s gifts. About 35% of conversations resulted in a purchase.

Chatbots are distinguished from the same mobile applications by the absence of the need to get used to and master the interface. Dialogue is dialogue, and even pensioners are ready to communicate with chatbots.

Where can problems arise here?

Most problems arise when the user has high expectations. It is worth informing the buyer from the very beginning what functions the bot can perform. Together with the developer, you need to adequately assess at what points in the dialogue you should use “buttons” in order to clearly understand the client. It is also necessary to work out algorithms for transferring to a live operator. The main thing is that the person does not feel that the bot cannot cope, and at the same time “does not allow him to talk” with a real person. This is the most disastrous situation.

Is a chatbot scary? Or funny?

We often develop chatbots for support departments. On last stage Employees often join in testing them contact center. Often they want to prove that the chatbot is ineffective because they are afraid of losing their job.

One day, support staff started asking the chatbot questions that they themselves didn’t know the answers to. There was a case when a CC operator, in testing mode, wrote to the chatbot “I hate you. You are too smart. You will deprive me of my job.”

Chatbot interlocutors often perceive them as real people. As a result, funny situations arise.

One day, a user complained to the bot about the site’s interface that he couldn’t find anything on it. The chatbot asked for forgiveness for this and received the answer: “What are you asking for forgiveness for, it’s not your fault.”

What will chatbots become in the near future?

By according to Oracle 80% of businesses plan to have chatbots by 2020. At Digital Intelligence, we see chatbots of the future as a universal interface to any system, which will largely replace modern website and application interfaces.

The chatbot will instantly process information from multiple sources and find the answer to your question. But the main trend of the future will be the development of artificial intelligence and machine learning, which will allow chatbots to very quickly find the most important information in any process. effective solution. This is precisely the concern of those e-commerce employees who believe that “bots will take our jobs.” We think there is no need to worry about this. Most likely, in the future, specialists of a different profile will simply be required - those who know how to set up bots and work with them in a single connection.

Guys, we put our soul into the site. Thank you for that
that you are discovering this beauty. Thanks for the inspiration and the goosebumps.
Join us on Facebook And In contact with

Robots, or simply bots, are small personal assistants on your gadget. They are programmed with numerous functions and are incredibly useful in the most different areas our life.

• @iVideoBot - This is the easiest way invented by mankind to download videos from YouTube. You just need to send the bot a link, select the format and size from the proposed options. And voila! Audio or video is already on your gadget.
• @utubebot And @ytaudiobot - 2 more ways to download materials from YouTube in one click.
• @SaveVideoBot- this bot can download videos from all other platforms, including even - attention! - Instagram.
• @auddbot- a bot that is an analogue Shazam apps. He guesses the song from the excerpt. All you have to do is send him voice message a few seconds of melody - and you will receive its cherished name.
• @ImageSearchBot will allow you to find any image. You just need to enter a word or phrase and select the quality.
• @joinstabot gets likes on Instagram. It works properly, although it’s a little unclear why it’s needed, but vanity lovers should keep in mind that if you suddenly increase more than 1,000 likes, your account can be blocked indefinitely.

• @topdf_bot- an incredibly useful and cool bot. It converts various files into PDF format - just send the file.
• @pollbot- you can easily conduct a survey or vote with this guy in any chat. Moreover, you write down the answer options yourself.
• @MyTeleCloudBot is a limitless cloud in Telegram. You can store and categorize absolutely any files. Isn't it a miracle?
• @temp_mail_bot- this assistant creates Mailbox for 10 minutes if you suddenly need to quickly register on some site.
• @voicybot is the perfect bot for the lazy and tired who can no longer type messages. You just need to dictate the message out loud, and he will give it to you in text form.
• @uber_promo_bot periodically sends promotional codes for Uber taxis.
• @storebot - This is a bot of bots. Will help you choose an assistant to suit every taste.

• @Chess_Bot- you can play chess with this bot.
• @saytextbot- This funny bot will convert your text message into an audio file. U male voice voice acting like in films - you can entertain your friends with such messages.
• @strangerbot arranges a chat with a randomly selected user of the same bot. Who knows, what if you meet your fate or good friend? Well, or you just have nothing to do.
• @PandaQuizBot is an entertaining quiz with more than 25 thousand questions. Good way pass the time in line.
• @zodiac_bot- if you believe in horoscopes, you should pay attention to this operational bot. Its developers guarantee, if not the veracity of the predictions, then stability and daily alerts for sure.
• ‎@PokerBot- poker bot. You won't make any money with it, but the game is quite gambling. Your opponents are 4 “random” players - users of this channel.

"Bots" are a new trend in social networks. This topic interests many users. why is it needed, how to create a “bot” on VK, its page, independently and without using any programming elements? The answers to these questions can be found in this article.

What is a "bot"?

In fact, everything is very simple. A “bot” in social networks is a non-existent person who in fact simply does not exist. It's fake. They are also called “dead souls”, just like in Gogol’s story, non-existent people, but listed in some structures. IN in this case fakes on the social network VKontakte will be considered. The “live” display of the “bot” is presented in the form of a profile of those people who actually do not exist, or these are abandoned user pages, and perhaps even copies, but the real owners do not know about it. What can such a “bot” do? Nothing, he just exists and that’s all. He can't read or write or anything at all.

What is a "bot" for?

As mentioned above, all such fakes do absolutely nothing, unless, of course, you install them special program. So why is it needed at all then? Its main function is to use such “bots” to increase the size of the desired group. To promote any community, subscribers or members are required. This is exactly the function that bots perform.

A search on the social network VKontakte shows in the first positions exactly the community with the largest number of people. This is a very important point, the more participants, the higher they are in the ranking. It turns out to be a kind of paradox. To invite real people, you need to promote the community using bots. Gradually, real, existing users will join the group. The "living activity" of the community will develop and become a natural process.

How to create a “bot” page on VK yourself?

1. Anytime search engine We find the registration page in “VK”. In the "name" column, write any name. In the “surname” line, some, preferably common, surname. Click "register".

3. After confirmation, a blank page will open for you to fill out.

Now, after registration, we move on to the main question, namely “How to create a bot in VK?” On the new user page, you must fill in all required fields. Specifically the following:

1. Basic: you have already filled in your first and last name. So, you need to specify the gender, Family status, date of birth, city, language. Next - grandparents, parents, brothers, sisters, children, grandchildren - add the ones you need. Click "save".

2. Contacts: you need to add your country, Skype, personal website. You don't have to add the last points.

3. Interests: you need to enter your activities. Specify as possible large quantity, interests (books, travel, humor, etc.). Next, fill in your favorite music, then movies, TV shows, books, games, quotes and write something “about yourself” and then save.

4. Next is education, indicate the school, educational establishments and save.

5. Career: you need to enter your place of study and work, you can have several, and also save it.

6. Service: male only.

7. Life position: you need to fill in - worldview, the main thing in life, the main thing in people, attitude towards smoking, attitude towards alcohol, source of inspiration, and save all this as well.

So, the basic data is filled in, let's move on to the next stage.

Stage two - adding photos

So, let’s move on to the second stage of solving the question “How to create a bot in VK?” Now you need to add an avatar and photos.

1. Find suitable photographs on the Internet in the “pictures” section. Download them to your computer and save them.

2. It is advisable to find several photographs of one person. In this case, you can create the appearance of a “live and real user”.

3. We select a medium-sized photo for the avatar; the face of our “bot” should be clearly visible on it. No animals or any extravagant photographs are allowed. Everything should be close to reality. Upload to the VK page and save.

4. Now add photos, the more, the better. They can be anything. The “bot” can be depicted on them either in the singular or with a company or with a group of people. You need to upload at least 4-5 photos. Let's move on to the next stage and find out how to create a "bot" in "VK", the final steps.

Reviving the "bot"

Now we need to add more vitality to the fake.

The third stage of solving the problem “How to create a “bot” in VK?” implies the most active actions. You need to upload several videos and some audio recordings to the page of the new fake. And also add friends. How to create a “bot” in “VK” without programs, so to speak, in manual mode? This will require an investment of your personal time. The fact is that you will have to add friends manually. You will have to create and accept friend requests daily. To make the fake user seem even more alive, he needs to communicate. "bot" in "VK"? You can do this in two ways: install a special program or write in the chat yourself every day and communicate with other users on his behalf.

As you can see for yourself, there is nothing complicated. However, it is worth taking into account that in order to create a full-fledged “bot”, you need to spend your free time.

Facebook bots, Telegram bots, Slack bots; they take away the jobs of call center operators, help order a taxi, download a movie, and for some they replace a psychologist

We asked the editor-in-chief of a technology site Apparatus Andrey Brodetsky explain what you need to know about bots - the most discussed (and perhaps very overheated) area in IT.

Why is everyone suddenly talking about bots?

The mobile application market is stagnating. In the AppStore, one and a half million applications compete for the attention of users, in Google Play- even more. At the same time, people are not exactly eager to try something new. Back in 2014, the majority of Americans did not download a single app (at the end of the month). And according to ComScore, users spend 80% of their time in just three apps. The top stores have been occupied by large players, and the costs of promoting most applications simply do not pay off. It has become almost impossible for beginners to get into users’ smartphones.

But there is a segment of this market that continues to grow actively - we are talking about messengers. WhatsApp recently passed the billion user mark. Facebook Messenger– 900 million, WeChat – 700 million. The total audience of the four most popular messengers back in November it overtook the most popular social networks.

In parallel with the decline of applications and the explosive growth of instant messengers, artificial intelligence technologies, natural language processing and cloud computing.

Bottom line: developers began to understand that not all tasks need to be written separate application. Instead, you can pester the user in those programs that he has already installed, and solve the same problems with with the help of the lung software (bot) built directly into the chat. Thus began the bot revolution.

What are bots?

Bots are programs that interact with the user through a conversational interface. For the user, it looks like a chat with an ordinary person (hence the other name - chatbot). Some people find this the most convenient this moment interface for communicating with a computer: you ask a question or report your need (“Where is the nearest Italian restaurant?”, “Is there any news about Apple?”, “I need a taxi,” “I need a gif “I’m sad”,” etc.). ), and the program gives you an answer, fulfills your order, or recommends you a gif, a link, a translation of an unfamiliar word, or whatever you want.

Eat different ways interaction with online services - you can place an order through a separate application or place it on the website, you can write or call the support service. Facebook Messenger Chief Developer David Marcus is confident that the best we can do is replace it all text communication in the messenger.

What can bots do?

Almost everything you need from Internet services. Booking a ticket, calling a taxi, choosing clothes, shopping, payments and transfers, search, weather, news, maps and routes, downloading movies and music - bots will help you with all this.

Theoretically, a bot can replace customer support any business. Sony Company Pictures already replaced 70 chatbot operators from the startup Msg.ai. Gradually, it will be the turn of government clerks - why spend the budget on them if a program can check and issue documents according to given rules? (Bot written by a nineteen-year-old Briton, helped Drivers could save $3 million by appealing incorrectly issued parking tickets.)

Tina from Iran is the most popular bot on Telegram. She has more than 2.6 million subscribers. Tina can report the latest news, send music on request, give advice, and just chat.

The Chinese chatbot Microsoft Xiaoice, imitating the behavior of a seventeen-year-old girl, is a real star of Chinese social networks. Xiaoice communicates with tens of millions of Chinese on WeChat and Weibo. She knows how to sympathize, joke, can be cunning and avoid conversation. Many correspond with her to brighten up loneliness, even knowing that this is a program. It sounds like an episode of Black Mirror, but it's reality. The developers call Xiaoice the largest Turing test in history. Her Japanese sister Rinna is less popular - she has only two million subscribers on Line (also a messenger). Users mostly chat with her about anime. And here is an experiment with Tay, an English-speaking Microsoft chatbot, did not work out: trolls from the anonymous forum 4chan in just a day forced Tay to repeat Nazi slogans, glorify Hitler and the Holocaust.

Silicon Valley startup X2AI has developed a psychotherapy bot to help Syrian refugees in Lebanon. There are more than a million forced migrants there, about 20% of them experience psychological problems. There are not enough doctors and volunteers in little Lebanon to help all these people. There is enough bot for everyone.

Where are the bots?

The first full-fledged messenger platform was Chinese WeChat. With multiple app integrations and payment options right inside WeChat messenger became incredibly popular. Twenty million Chinese companies operate in it in an “app-within-an-app” format through so-called official accounts.

The first Western messenger to discover the potential of bots was Kik , founded by Canadian Ted Livingston. It has 275 million users, mainly in the US and Canada. Kik is extremely popular among teenagers - 40% of American teenagers use it. Livingston believes that the future of instant messengers lies in bots, not artificial intelligence: It's all about speed, not complexity, he says. The main owners of Kik bots are large brands and fast food chains. They use simple automated bots without human speech processing - mainly for commerce. At the end of March, Kik opened its bot store and API for their development.

Messenger Telegram, founded by Pavel Durov, opened a platform for creating bots last year, and recently updated it. It is already active thousands of bots. Bot pirate library Flibusta helps you find any book and send it directly to your phone. The blocked RuTracker bot searches for and downloads the torrent. With the help of bots in Telegram, you can find out the weather and news, play games, translate words, download movies, and add pictures, gifs and music directly to the conversation.

At the end of March, its platform for bot developers Microsoft. The company lost the battle for mobile applications to competitors from Google and Apple and is now trying not to miss new stage,placing a bet on smart bots with integration into mail, Office, Skype and other services.

Stands apart corporate messenger Slack. Manager bots running in Slack can schedule and facilitate meetings, monitor task completion, abstract important discussions, automate workflows, and order food directly from chat. Slack can be configured for more than just work tasks. One Swedish programmer integrated Slack into family life and with its help manages the family calendar, monitors the movements of children and buys groceries.

The most awaited event of the year was the opening platforms Facebook for bots. Now any developer can create a bot for their business. I tested the first bots available in the Facebook store. The cute cat Poncho told me the weather forecast in my city, CNN, after two iterations, began to produce news relevant to my interests, and the Spring online store helped me choose inexpensive sneakers in a few clicks. (Too bad this store is in New York.)

Why are bots bad?

Many problems are more conveniently solved through GUI site or mobile application than through the bot's step-by-step text interface. Pressing a few buttons (for example, choosing food from a restaurant menu) is faster than typing on a keyboard. Developers are trying to solve this problem using hybrid interfaces that combine graphical interactive elements(buttons, links, photos) with the usual text chat. But even with this approach, using the bot can be confusing. (Testing many bots in Telegram, I began to get confused in the tree of options after several iterations. Where did I go? How to go back? At such moments, I want to return to the clear graphical interface of the site, where all the navigation is visible.)

Chat with a bot not very comfortable psychologically. Despite all the impressive advances in natural speech processing, we still have not taught a computer to understand it one hundred percent. Entrepreneur and investor Andrey Khorsev(908.vc) believes that artificial intelligence is now in a state of “uncanny valley” - when we ask a question, we are not sure that we will get the right answer and that our question will be correctly understood. Inappropriate use of a company's text interface can even destroy a product due to the negative effect it creates.

Besides, not all applications can be replaced by bots. Even this standard process how to buy an airline ticket turned out to be too complex for a text interface, says the executive mobile development AVIASALES Ivan Kozlov. The company’s attempts to sell tickets through bots were unsuccessful - people still find it more convenient to use an application for this.

How will the bot market develop?

Bots have become so hot topic, that everyone rushed to do them. The excitement of developers is fueled by daily news about millions of investments in startups working on artificial intelligence. Silicon Valley investors joke that the share of pitches about startups related to bots reaches one hundred percent.

But the bot market may become saturated even faster than the app market, since bots are an order of magnitude easier to develop. The winners will be those who were among the pioneers and those who can offer the user something incredibly convenient. Much depends on how the bot ecosystem develops. Obviously, the most popular will be bot stores owned by the platforms themselves, that is, instant messengers. Horsev predicts: “Everything will develop much faster than with app stores.”

“The problem of promoting bots is acute now,” says investor and VP for business at VoxImplant Viktor Koch. – I think that in six months bot engine optimization will appear, and with it BEO specialists (similar to SEO - search optimization ) ».

What will happen next?

Bots likely won't take over the entire Web and kill applications, as headlines promise. But they will definitely take away the work of many support staff and operators taking orders. E-commerce, information services– in these areas, bots will take away a significant portion of traffic from sites and applications. Just as applications did not kill, but complemented the Network, bots will not take over all the functionality of sites and applications and will instead complement them. The niche that bots will occupy may turn out to be very modest. “Bots can repeat the path smart watch“- says Kozlov from AVIASALES.

The most interesting and hard to predict question is: what can you do with bots that couldn’t be done with a website or app? Perhaps the gold mine is here, and not already in transfer existing models V new channel. In any case, the move from apps to bots seems less revolutionary than the move from web to apps.