On the Internet, the HYIP around chatbots (in particular, Telegram) does not stop due to media noise, undeniable advantages of the platform, promotion policies, development tools, etc.

You watch the news: well
, there is no life without chat bots!
But if they are not there, the trains will go off the rails, the planes will fall, people will die from anguish when they cannot find pictures with cats.
But let's put a hand on the heart : when was the last time you ordered something in the online store
through a chat bot ?
')
Who are all these people who order the development of bots for their stores?
Typical chat bot of Vasya Limited:
>> automates the
flow of waterfall applications from
5 people a day ,
>> merges 4 of 5 applications, obtained in blood of Yandex Direct,
>> if you are lucky, a person will find a phone number and call,
>> but, most likely, "Uh, where to press?", and then close and go google further.
What does the
owner do when sales are “automated”:
>> orders in excel-table
>> fills out mailing forms on parcels
>> standing in line at the mail with a bunch of parcels (every day!)
>> enters the track numbers in the excel-table, then sends to customers
Maybe it’s enough to build “technology” out of the blue where people really need it, while people are busy with the routine of robots?
Why bots do not gain the expected popularity?
- They do not solve a real problem much better than current tools.
- They do not solve a real problem much better than current tools.
I.e
- The problem that he solves is not really: for example, to report the current color of the sky or the distance to the Sun
- The problem is there, it is solved, but it is solved only slightly better than current decisions: to set a trend, it’s not enough “just better” - it should be much better , since switching to something new is always energy-consuming. The gain from the new should, obviously, cover the loss on the development of a new technology.
Why is this happening and what to do?
- Even just to see a real problem in the world around us is not an easy task.
The fact is that in the world
all needs , one way or another, are
already satisfied . Most decisions are something that we take for granted.
We cannot imagine that this would be implemented differently, we do not even think about it and do not see any problem at all.
For example, when there were no trains, moving around the world on horseback took months, years. But people did not see any problem in this: in their picture of the world it could not be otherwise. Before the advent of the iPhone, no one saw the problem in the physical buttons.
- Most developers are focused on the process, not on solving the problem.
The motivation of developers is getting new knowledge and skills, not solving problems. If the solution is not at the edge of the possibilities, it does not even deserve attention, no matter how effectively the problem is solved, it is important that the solution be cool.
Entrepreneurs, even if they see a real problem, do not have enough knowledge and skills to come up with the concept of a solution from current technologies. Therefore, it is often just hammered into finding a solution, content with what they offer.
As a result, programmers are playing with new technologies, trying to stick them wherever possible.
And businessmen have to rummage in this pile of garbage. Sometimes suitable products suddenly appear from this pile.
Although it is enough just to rely on what we already know and solve more mundane tasks.
Real case without fantasies
Our company is engaged in the production and sale of electronic devices. Customers around the world. Every day, 5-10 parcels are sent (by the standards of online stores, quite a bit) by mail, by transport companies; abroad, across Russia.
- During the day come orders that fall into the database - table google sheets.
- At the end of the day the packages are packaged, the addresses on the packages are filled.
- Parcels are taken to the central post office (it operates around the clock), they are sent.
- Checks are photographed and sent to a special Telegram-chat.
- Someone from the staff interrupts the track numbers from the chat to the table.
- Track numbers are sent to customers by email; letters in Russian and English.
It seems to be nothing unbearable in this.
But, if you count the time that these small routine operations take in total,
it turns out that about 2 hours each day is spent on:
- Filling addresses on packages
- Breaking track numbers into the database
- Checking track numbers (sometimes numbers are poorly printed)
- Mailing letters to customers in different languages
- Monitoring of other parcels: have you not been stuck at customs, have you been in the mail of the recipient for a week, etc.
People are expensive

- If earlier it was possible without thinking to hire a girl, “bring it up, feed it, dump it, don't interfere” for 15k a month, then the current conditions do not forgive inefficiency.
- First of all, biorobots in small business are expensive because they often take turns, require a lot of attention and control, training, etc. All this takes time from the key people of the team.
- Each new employee in a small business significantly increases management costs, as there is no system.
If the process is formalized, it can be automated.
We tried to implement various systems such as 1C, bitrix and so on. Not “logged in” for a simple reason - applications load for a long time; in order to make, for example, a waste of 300r, you need to wait for the application to load, poke in several fields, put labels, etc.
As a result, expenses, tasks and other information are simply not entered.
The system with telegram-chats "got accustomed": it is always at hand, it loads quickly, cross-platform.
Conclusion: the main advantage that Telegram gives for small business is the
speed of information input ,
everything that takes more than a couple of seconds and a couple of clicks will not be done in the end .
OK, we’ve added the information, but, in fact, all these chats are just a dump of raw information: receipts, expenses, purchases, contacts, checks, etc.
And here we logically come to the fact that we need a bot that would process this information - the main thing is that the information is entered.
Recognize track numbers
The task is minimum: let the bot at least recognize the track numbers from checks, check the checksums and give it to the chat, so that it remains just to copy and paste the letters into the template, and not to interrupt manually.
The problem is that there is no track number in the form of a bar code or QR code on the postal check
And then
ABBYY CLI OCR for Linux comes to our rescue
.
I will tell you more about this tool in another article: ABBYY text recognition with a bunch of settings, languages, barcode and QR codes, works an order of magnitude better than Google.
We feed our picture to it, at the output we get a sufficiently recognized text, we need to parse it, clean it, find the track number in it, check and correct errors (for example, the number 8 is taken as 0).
Well, since we pulled a track number from the check, we can send it automatically, too, the only question is to whom? You need to somehow match this track number with the corresponding order. We make a small service on php, into which we transfer our database from Google sheets.
We print address blanks (which are pasted on the package instead of hand-inscribed), a pick list and a back with a QR code with reference to the order.

After sending the packages directly at the post office, we photograph the checks attached to the stubs with the QR code and discard the bot. Everything, then everything happens automatically.
- Recognized track number of any delivery service.
- Checksum is checked.
- Determined by the type of departure and language
- An email is sent to the client with the track number in the desired language.
- The track number is entered into the database, the order status changes
- The fate of the parcel is automatically tracked.
- When delayed, notify us for action
- Upon arrival, the package reminds the customer to receive it.
- After receiving, after 2 weeks, ask for feedback
The system saves time, eliminates the routine and human error, increases service and, most importantly, profit.
Thank you for your attention to everyone who has read my soul cry to the end. I hope I was able to convey my vision of what tasks chatbots really need to solve. Formalized tasks should be solved by machines, people should be engaged in creativity.
Stop doing newfangled crap that does not benefit (around a bunch of similar tasks) let's make the world a better place)
PS If the topic is interesting, I can write in more detail. There are also many ideas and prototypes of such systems, which do not have enough time to bring to mind; if someone has a desire to participate - well.