Hello, Habre! My name is Svitlana Ivanova, I am the team’s IT recruiter X5 Digital.

I want to share the results of an experiment in which my team and I automated the process of writing vacancies and compiling Boolean queries using AI.

Previously, we learned how to write personal letters to candidates faster using a letter generator based on ChatGPT. We did it with help podbor.io (platforms for finding IT specialists). I described this experience in detail here. In short: we succeeded. We reduced the time spent on writing personalized letters and were inspired by the decision to test AI tools on other routine processes. After all, volumes are growing, but the number of hours per day is not increasing. More automation of the god of automation!

In the experiment that I want to talk about in this article, we formulated several hypotheses and selected the two most priority ones for us:

1. If AI is good with letters, will it be good with Boolean queries? Recruiters spend a lot of time building complex queries that have different operators and keywords to filter search results. Can we significantly reduce this time without losing quality?

2. II, speed up the process of editing job descriptions for us, please! A well-written vacancy is an important part of attracting the best candidates. In a sea of ​​identical vacancies, it becomes more and more difficult to keep the candidate’s attention. At the same time, recruiters have little time to write catchy job descriptions without using cliched phrases. Will ChatGPT be able to speed up the process of writing a job description, which the candidate will not pass by?

So, we checked.

Case #1. We learned how to write Boolean queries using a chatbot based on ChatGPT

If you have not used Boolean queries before: *Boolean search – a search method that uses logical operators to obtain more relevant search results on websites and social networks.

Many recruiters are aware of the existence of this tool, but not everyone is familiar with its capabilities or may make mistakes when using it. We decided to check how a chatbot based on ChatGPT writes such requests.

Examples of Boolean search queries can include the use of “AND”, “OR” and “NOT” operators in combination with keywords and phrases. Example:

• “software developer” AND “work experience of at least three years”

• “sales manager” OR “marketing specialist”

• “engineer” NOT “trainee”

We used an AI-based chatbot to which we asked “write a resume search query to HH.ru. DBA with PostgreSQL, Linux (Ubuntu), Ansible, Bash, Shell (scripts):

Result:

Chat can write Boolean queries to various resources, we tested several. Here are examples of issuing a request for LinkedIn and Zen:

Result:

Result:

It turns out that AI coped with this task quite successfully.

Case #2. We learned how to write job vacancies using a chatbot based on ChatGPT

A well-written vacancy is the key to attracting the best candidates. Often, recruiters have little time to write catchy job vacancies without clichéd phrases. We tried to create new vacancies using a Chatbot based on ChatGPT.

You work, you work, and here – bang! – you receive a vacancy from the head of the development department with a short description: we are looking for a kind, energetic and cheerful person who knows the JavaScript language perfectly, like his native language, and has a passion for traveling. We need a real adventurer!

The customer does not have time to delve into the details of the vacancy. From these introductions, you should write an attractive vacancy announcement, and do it quickly, because the vacancy needs to be closed “yesterday”. Has this happened?

After the AI ​​tool handled the Boolean queries, we decided to try handing the process over to it.

Through chat, we can create job templates and receive generated text based on input parameters and requirements.

For example, you can set the following parameters:

• Job Title: Web Developer

• Requirements: knowledge of HTML, CSS, JavaScript

• Work experience: from two years

• Responsibilities: development and support of web applications

Based on these parameters, the chatbot generated a job description, which we can edit and refine at our discretion. This saves time and simplifies the process of creating job descriptions from scratch.

What conclusions did we make:

• In these cases, the main goal was achieved – the automation of writing vacancies and compiling Boolean queries.

• Often the bot wrote Boolean queries, where the output was zero. But by gradually simplifying Boolean queries, we expanded the output gap and thus saved time.

• The chat will write the text for you, but keep in mind that you will still have to edit it.

• The more introductions you give to the bot, the more beautiful and correct your desired text will be.

• We will definitely continue to use AI tools in our processes (and will be happy to share our experience with you later).