AI Interview in the HeadHunter Chat: Screening Without Calls

A candidate has applied to a vacancy on hh.kz, and from here on whoever starts the conversation first wins. The recruiter opens the application chat, writes a greeting, asks the same questions about schedule and experience as yesterday, and repeats this for dozens of applications. This can be automated: HRStream connects to HeadHunter and runs the first interview right in the hh application chat, following the recruiter's script, then hands the recruiter a ready summary. The candidate does not have to move to WhatsApp or leave a phone number: they reply where they applied. Below we break down how it works.

Why silence in the chat costs you candidates

According to an hh.ru survey of 2.2 thousand professionals (October 2023), 40% of job seekers lose interest in a vacancy if the employer does not reply within three days (source in Russian), and 90% would rather get a rejection straight away than be left without an answer. The survey covered the Russian market and its methodology is not disclosed, but the picture is familiar in Kazakhstan too: whoever replies fast gets the candidate.

Earlier, hh also wrote that more than 40% of job seekers wait for an employer's reply for only two or three days (2021, source in Russian). For high-volume positions the window is even shorter: an application is a single click, and the candidate is applying to neighboring vacancies at the same time.

What hh itself offers and where an ATS is needed

HeadHunter has its own candidate chat and built-in tools: on the page about auto-replies and the chatbot (in Russian), hh describes how to set up automatic application processing and a bot in the "Applications" section. That is a good base for simple cases: a greeting, a template rejection, a short questionnaire.

When there are many applications and the conversation has to be substantive, other tasks appear:

• run a dialogue following a script that differs for a sales assistant, a courier and a warehouse worker;
• answer the candidate's questions about schedule, terms and address, not only ask your own;
• collect the interview result into the candidate card and move the candidate through the funnel without manual transfer;
• see candidates from hh, WhatsApp, OLX and other channels in one place.

This is exactly why HRStream connects the hh application chat as one more AI interview channel. We covered the connection to HeadHunter and application processing in the articles on the HeadHunter integration and setting it up in 5 minutes.

How it works: from application to interview result

Step 1. The application enters the funnel

Applications from hh are imported into the vacancy funnel. If you have AI resume scoring against a candidate profile and funnel rules set up, they run first. The chat interview is the next step, not a replacement for resume screening. Read more in the article on AI candidate screening.

Step 2. The bot writes in the application chat

At the "Launch" stage the bot opens a dialogue in the application chat on hh and runs the interview following your vacancy script. The candidate does not need to look for you in a messenger or share a phone number: they reply in the familiar hh chat.

Step 3. A scripted dialogue

The bot asks the script questions: experience, readiness for the schedule, district, start date, salary expectations. You set the script up once per vacancy. It can be generated with AI and refined by hand; how this works is described in the article on the HRStream web chat. The dialogue runs in Russian or Kazakh.

Step 4. Answers to the candidate's questions

Candidates ask questions too: how much is the pay, what is the schedule, where is the office. The bot answers such questions from the vacancy knowledge base and, where needed, from organization data. If the knowledge base has no answer, it honestly says it will check with the recruiter, and the recruiter gets a notification.

Step 5. Reminders

If the candidate goes silent, the bot can send a reminder. The reminder frequency is set in the script.

What the recruiter sees and how a human steps in

The whole hh conversation is shown in a single candidate card, next to the conversation from other channels. The recruiter replies right from the card, and the message goes to the same application chat on hh.

• If the recruiter replies to the candidate from the HeadHunter account, the bot is paused and the recruiter gets a notification.
• If the candidate writes "stop", the bot stops communicating on this vacancy.
• One interview is held per "candidate and vacancy" pair, even if the person reached you through several channels.
• To a direct question "Are you a bot?", the bot honestly answers that it is an AI assistant.

The HeadHunter integration settings show the status of the chat channel, so it is easy to tell whether everything works and whether you need to reconnect.

Interview result and moving through the funnel

After the dialogue, HRStream produces a score and a short summary with quotes from the candidate's answers. The outcomes are: the candidate passed, was rejected, did not finish the interview, or the dialogue needs to be handed to a recruiter.

Next, the auto-advance rules that you define yourself come into play: those who passed can be moved to the interview stage, and rejections can be sent as rejections. Candidates who never replied are not moved by the system under these rules. The "Rejected", "Offer made" and "Started work" statuses can be synced back to HeadHunter if this is enabled in the vacancy settings.

The hh chat and other channels

For a company this means one thing: the interview is not tied to a platform. Wherever the candidate comes from, they go through the same script, and the result lands in one funnel. The overall picture of channels is described in the article on the HRStream omnichannel messenger.

A high-volume hiring scenario

Say you are hiring sales assistants and several dozen applications arrive from hh every day. A working setup looks like this:

1. The application is imported, and AI scores the resume against the candidate profile.
2. Suitable candidates move to the "Launch" stage, and the bot writes to them in the application chat.
3. In a few minutes of dialogue the bot finds out experience, schedule, district of residence and start date, and answers questions about the terms.
4. Those who pass the qualification move to the interview stage with the recruiter, the rest receive a rejection.
5. The recruiter starts the day not with correspondence but with a ready list of people worth inviting.

How this shortens the time to close a vacancy and which metrics to watch is covered in the material on time-to-hire.

Launch checklist

1. Connect HeadHunter in the HRStream settings and make sure the chat channel is active.
2. Enable the AI interview in the organization and on the vacancy.
3. Prepare the script: five to seven questions for the role, plus expert questions if needed.
4. Fill in the vacancy knowledge base: pay, schedule, address, duties, frequent questions.
5. Set up the auto-advance rules: who goes to the interview, who gets a rejection.
6. Test with a trial application, and after a week refine the script based on real dialogues.

Frequently asked questions