How chatbots save business resources and speed up work, using the example of POSiFLORA

Chatbots help businesses serve customers: they answer questions, send emails, and can make an appointment with a specialist. In addition to helping technical support, chatbots automate business processes. They train staff, accept applications, and provide access to data. In this article, we describe how OrbitSoft developed chatbots for POSiFLORA, and what tasks they solved.

POSiFLORA develops and sells software for the floristry business. With one program, flower shops can place orders, plan purchases, build work with employees, analyze sales performance, and plan development. We have previously talked about the POSiFLORA project in other articles.

How POSiFLORA helps the flower business

If the program crashes, flower shops and salons contact technical support. To help solve the issue, technical support needs data regarding a customer’s device: e.g., tablet model, license expiration date, and current version of the program.

To get the data, a technical support specialist creates a request for the internal monitoring system: they enter commands into the interface, the system analyzes these commands and displays the data. Because of this, technical support response can be slow, and a client can become frustrated.

The management of POSiFLORA wants to keep track of sales made through the program: e.g., how many flower shops and stores are registered in the system, how many actively use it, and how many invoices have been paid during a certain period.

To reveal this data, management set a separate task, and managers manually collected data for a report. Creating a report was not instantaneous, and took some time.

Both tasks could be solved if a way was found to automate the collection of data from the system. To do this, OrbitSoft developers decided to create chatbots. Chatbots work like a command line on servers, but they are faster and more efficient: in just a couple of seconds, one command performs up to 40 actions and returns a result to the user.

You can use ready-made public chatbots from popular online services, or you can create a chatbot from scratch. We believe that custom chatbots are better for this work, and here’s why…

Data security. Custom chatbots are located on the client’s servers and allow you to restrict access to data. They don’t depend on the platform and the stability of its work. A chatbot in an online service will forever remain in the bot database. They are easy to find through search or BotFather, whereupon one can access their data.

Unlimited feature set. Unlike the standard set of chatbot functions from online services, custom chatbots are developed specifically for business tasks, and can perform highly specialized functions. For example, they help manage accounts, make calculations, and respond to customer inquiries.

Possibility for improvement. The customer can change, add, or update a bot as needed.

Official software, community, and support. Public bots are built on developer frameworks that can be updated or closed. We use the official API, open-source codes and programming languages, all with the Apache Free Software License.

The tech support bot reduced data collection time to 15 seconds. To speed up the collection of device data for tech support, OrbitSoft developed a chatbot. In 24/7 mode, the hotline employees ask the chatbot for customer indicators, such as information about status of the terminal, and uploading for recalculating data.

Previously, the interface issued data in five minutes. Now the request processing time has been reduced to just 15 seconds. The bot saves time and simplifies work for tech support. You can communicate with the bot using commands or through an interface. POSiFLORA tech support uses query with command, and it saves time.

For example, the user enters the /get_evo command and the bot generates a report regarding which of the clients has recently used Evotor, a third-party cash register software.

There may be complex queries, for example:

/update_b 2731 31286ef4−6f50−4c88-b8cf-c9b726319b3f

This command tells the system it needs to update information on the client with an identifier 31286ef4−6f50−4c88-b8cf-c9b726319b3f by object b 2731.

The chatbot understands that the object is a bouquet and recalculates statistics on bouquets for the required period for the client.

A reporting bot saves management time when contacting managers.
The developers have integrated the chatbot with the POSiFLORA API. Now the bot can create reports on the movement of money, paid bills, or active users on the site.

For management and customers, communication with a chatbot should be logical and intuitive, so instead of commands, management uses a clear interface with built-in menus or buttons. Now management doesn’t need to set tasks for its employees to create reports. They can get all the necessary information on their own from the chatbot.

Building chatbots in Python

Python is the most popular language for building bots. It’s powerful, fast, has a wide range of libraries, and Telegram API support.

Often used for PHP and Java development.

We use Telegram API

Telegram is not the most widespread, but is the fastest growing platform in Russia, with a solvent and active audience. For example, more people use WhatsApp, but not everyone there is ready is ready to use chatbots. Also, WhatsApp doesn’t have an official API.

It will take at least a week to create a bot.

It will take a week and a developer to create the interaction of a simple chatbot with the Telegram client.

Complex bots require two developers. The first creates the interaction with the Telegram API, and the second describes microservices that perform internal work in the backend.

Required capacity for chatbots depends on the task

A chatbot can be used even on a home computer if there is a hard drive, at least two cores, and at least 2 GB of RAM.

We recommend using a server in a data center, a minimal cloud solution where you can pay for hours of service.

Additional RAM is needed for capacious tasks, powerful calculations, or for converting video to mp4.