Since when the bots appeared in Telegram, they have won many messengers and social networks, having turned from innovation into a fashionable trend. Andrey Khorsyov, one of the creators of the bot that collects public data in the Ukrainian registries of Opendatabot, in his article discusses why among so many bots there are so few good and useful ones. The text originally appeared on Facebook, the editorial publishes it with the permission of the author.
Why is there no good bots in the world?
A year and a half ago, in the California Today Show, we announced the launch of Botagedon, a worldwide trend in the creation of bots. We must admit that the mass appearance of useful bots has not started.
Of course, in our laboratory there is an Opendatabot, which grows at a rate of 10% per month and has 52,000 users, 15,600 of which are subscribed to various services. But this is an exception to the rules.
Let's look at the rules that the bots are working on.
1. AI Bots
AI-bots, the first futuristic trend born on the hip-up "bots + AI (artificial intelligence). The hypothesis was that the AI-bots would speak to us with human language, and the first thing they wanted to do was talk to their dead relatives and create a seller who would sell anything, whipping away the buyer.
In reality, AI is so weak that it is confused in the three pines, unable to support the context of the conversation on the narrowest topic. This means that the AI-bot can not at the moment sell even one model of the phone, no matter how much its neural network was trained, not to mention the whole store of goods.
2. Push Button Bots
As a reaction to the complexity of creating AI, a buttonbot paradigm has arisen, where all the answers are written by writers and the user wanders around a certain graph of decisions. A good example is Christian Gray Chatbot with a well-written script that creates a sense of dialogue with AI.
The problem with this approach is the lack of good writers and the still obscure business model.
3. Bots-shapes
In the environment of lawyers, bots provoked a fair interest, and the first thing they began to do was to reduce their routine work, in particular filling out various forms, for example, a statement about the return of goods to the store.
But even in the form of a bot people do not like to fill out forms, and this conversation can not be made fun, in part because of the lack of good scripts.
4. Showcase Bots
Zuckerberg's pink dream, a bot showing a product card with the "Buy for $ 19.99" buttons. In fact, on this functional all the first version of the messenger platform is built.
These bots do not give any new experience and are like obtrusive sellers. In Facebook for a year understood the sadness of this direction and began to turn the platform in the direction of entertainment. Because before you sell, you need to entertain the customer.
5. Bots search engines
The interface to the database in the form of a bot - asked a question, got an answer. In fact, the repetition of the Google paradigm, but in the form of a window messenger.
The questionable idea of re-inventing the search string itself is meaningless, since the search engine paradigm is an established model of access to information.
6. Bots-payment systems
A bot that transfers money from your card to another user.
A special kind of perversion. With all the variety of means of payment, it is absolutely unclear why to pay a bot. The entry threshold in this process is quite high - you will have to tie the card.
Most messengers will introduce a money transfer service into the core of their system and are likely to simply destroy other types of transfers between users. And the most important question is where is the added value? In fact, you are not ready to pay for a transaction in a bot, well, except for an anonymous purchase of cocaine for bitcoins. But this is a messenger, you are authorized in it, what kind of anonymity is there?
How to grow a good bot?
The answer to this question lies in the limitations imposed by the bots paradigm.
What does the bot give us out of the box?
This is an authorized user and his alerts with a high degree of openability.
A good bot scenario is an alert that is configured by a specific user.
To grow a good bot, you need to have the information the user needs, and give him the opportunity to configure the receipt of this information.
What does this mean in practice?
- Bots for standing in electronic queues (for obtaining a passport, visa, buying a ticket in the case of withdrawing armor, etc.).
For the owner of the product it is an opportunity to ensure the full occupancy of the order system. For example, if you have all the chairs in the barbershop busy and suddenly someone did not come to the reserve, your bot can immediately write to those guys who tried to book a haircut, but there were no seats.
- Bots of personal alerts.
For example, you own company A. Every time when they write about you on Twitter, news or, God forbid, in the court registry - the bot will inform you about it.
Opendatabot works on this principle, tracing dozens of sources for changes in registration data of your companies, formation of a tax debt, sanctions, etc.
To track railway tickets there is a railwaybot, for standing in line for a biometric passport igov_bot. There are bots tracking the parcels of the "New Mail". Technically, you can create a bot for Ukrposhta, InTime and Diliveri, they all have an API.
Post created by @hrytsenko