"A productivity tool, not a judge": what AI really does in Ukrainian IT recruitment. We asked recruiters from SoftServe, EPAM, N-iX, Ciklum, Railsware, Levi9, PlantIn and others

Routine is the first thing given to the machine

If we generalize the responses from all companies, a clear pattern emerges: AI has entered where there is the most «mechanical» work and the least need for human judgment. Sourcing, resume screening, Boolean queries, meeting summaries, letter drafts, and job descriptions are the first line of automation that almost everyone starts with.

The entry point is well illustrated by the experience of Capgemini Engineering. «Copilot helps to form Boolean queries, resume screens, draft letters, and structure notes. Plus, in LinkedIn Recruiter, we often use their AI features: they help to form search queries faster and offer relevant profiles,» notes Olga Simonenko, TA Lead of the company.

This is a classic first step — automating point operations that do not require assessment, but take up a disproportionate amount of time. The next level is integrating AI into the entire incoming flow of candidates. This is the path that Master of Code Global is taking, where they are currently testing an internal bot for processing inbound applications: it answers requests, collects basic information and structures profiles before the recruiter opens the resume. In parallel, AI summarizes meetings with hiring managers and prepares qualifying questions for a specific role.

«This allows us to standardize processes, reduce manual work, and ensure more consistent candidate evaluation, leaving recruiters more time for quality communication and deeper interviews,» explains Natalia Yerisova, Head of Global Recruitment at Master of Code Global .

While Capgemini and Master of Code focus primarily on the entry funnel, Intetics has gone further along the entire chain: here AI generates match scores, standardizes candidate profiles to the client’s corporate format, generates interview questions and assessment forms. In essence, AI accompanies the candidate from the first contact to the final pitch to the manager.

Interestingly, even where automation enters the assessment area, companies leave a human as the verifier. In EPAM, for example, the initial English level check is carried out by an AI agent — but the final conclusion about the level is always made by a specialist. «The initial English test can be carried out by an AI agent, and the EPAM specialist checks the results and determines the level of language proficiency,» he says. Olga Makarova, Manager of the Talent Search Department at EPAM Ukraine. This is a telling detail: even the most advanced automation cases retain a person in the hiring process not as a formality, but as a meaningful filter.

AI notetaker: a small revolution in the interview format

It is worth highlighting a trend that almost all companies mention — AI notetaker during calls. This is probably the most widely adopted AI tool in hiring: it does not make decisions, does not evaluate the candidate, but significantly changes the dynamics of the conversation — the recruiter is no longer torn between listening and recording.

For Railsware, this is currently the only way the company uses AI in direct work with candidates. «It helps recruiters be more engaged and focused on the conversation, and also gives them the opportunity to review full structured notes after the call for deeper analysis,» explains Ulyana Tereshchenko, Senior Talent Manager.

Transparency is key here: Railsware always warns candidates and asks for permission at the beginning of the call. The same practice is at N-iX, where any use of AI during an interview is done only with the prior consent of all participants, as required by the GDPR. This norm is gradually becoming the standard for responsible companies.

At the same time, EVO records an interesting behavior: despite the fact that a candidate can refuse to register, in practice this rarely happens. «Such cases are rare, because it simplifies the process and can reduce the number of stages,» — Olena Oleksienko, Chief People Team at EVO.

In fact, AI notetaker has become the de facto standard — a tool that most candidates don’t refuse because it’s in their best interest. And for companies, it’s also a way to provide quality feedback: detailed notes allow the hiring manager to review the interview without having to call back.

Where to draw the line: AI as an assistant, not a judge

The most interesting thing about all the companies' answers is not that they use AI, but where they deliberately do not use it. And here the positions are almost unanimous: the final decision always belongs to a person. And this is not just a declaration — there is a very specific logic behind it.

Levi9, for example, uses AI to prioritize resumes — especially when a vacancy attracts hundreds of applications. But then — stop. «We do not delegate decision-making to AI. Recruiters still review each resume manually and independently decide what the next steps in communicating with the candidate will be. For us, it is primarily a tool for optimizing the process and prioritizing feedback,» — Tetyana Lobetska, Talent Manager at Levi9.

Capgemini Engineering formulates the same idea as a principle: AI is «a support tool, not a replacement for expert analysis.» In the spirit of this approach, Ciklum specifically emphasizes a boundary that it does not cross even with automated profile processing: the company does not use AI for direct communication with candidates. Although such cases are becoming more and more common in the world — bots at the first interview are no longer uncommon — for Ciklum, human contact in the assessment process remains fundamental. Similarly, N-iX: live people — not bots — communicate with candidates at all stages.

The common denominator of all these positions was accurately stated by SoftServe: AI is being brought into hiring «primarily as a productivity tool, not as an automated judge of candidates.» This is perhaps the most accurate description of the consensus the market has reached.

A meaningful rejection of AI in screening in Railsware

Against the backdrop of the general enthusiasm for AI automation, Railsware’s position looks particularly clear — and deserves special attention. The company deliberately does not use AI to screen and evaluate candidates. And this is not conservatism, but a principled position. Ulyana Tereshchenko, Senior Talent Manager at Railsware, clarifies: «Railsware assumes that we hire not „titles“, not „resume“, but people with their full range of knowledge, skills, experience, values, etc. Accordingly, getting to know a person is more than keywords on a resume. That is why involving a person is necessary in order to see the potential and more broadly assess the candidate’s background, especially due to the variety of job titles and experience that may be hidden behind them.»

The company also warns against over-confidence in AI solutions: «In any case, the decision must be validated by a person to prevent errors and biased decisions, which AI still „sins“ with.»

Data Security: GDPR and Internal Policies

Another common focus is the protection of candidates' personal data. This topic is especially relevant when it comes to transferring information to external AI services.

N-iX explicitly states that any use of AI during interviews is «strictly within the framework of GDPR policies» and only with the candidate’s prior consent.

Ciklum also emphasizes: «personal data of candidates and clients is not processed by artificial intelligence, and its use always occurs within the framework of internal data protection policies.»

This is an important signal for candidates: responsible companies do not «dump» resumes into any available ChatGPT — they have clear internal regulations regarding what data can be processed and in which tools.

A systems approach with ATS at SoftServe

Oleksandra Tolokh, Talent Operations Director at SoftServe, describes the most systematic approach among all the companies surveyed. SoftServe uses an internal ATS with elements of AI functionality, which helps to formulate position descriptions more correctly, parses CVs and matches candidates with relevant vacancies.

At the same time, the company clearly distinguishes where AI is appropriate and where it is not. «AI is partially involved in our hiring process — primarily as a tool to increase productivity, and not as an automatic „judge“ of candidates. Recruiters and sourcers use AI to better analyze the vacancy, understand the request more deeply, form a sourcing strategy, improve communication and conduct basic analytics of the results,» Softserve explains.

The phrase «a productivity tool, not a judge» is perhaps the most accurate description of the consensus the market has reached.

PlantIn: AI at the start of a new job

Natalia Mudrak, Talent Acquisition Lead at PlantIn, shares an interesting case of using AI at the early stage of working with a vacancy: «When we start working on a new vacancy, AI helps us with market analysis and building a realistic portrait of the candidate.»

This is a non-trivial application: not just finding people for a ready-made profile, but forming the profile itself taking into account the real market situation. In addition, the company uses AI to generate queries for LinkedIn and ideas for outreach, as well as for interview notes and structuring feedback.

What AI can’t really do

Despite widespread adoption, all surveyed companies agree on one thing: there are things that AI either can’t do or can’t be trusted without verification. And it’s not just about technical limitations.

Railsware formulates this boundary most sharply. Ulyana Tereshchenko, Senior Talent Manager, says: «Ultimately, hiring is about human-to-human communication. For us, those things that AI cannot yet (and probably will not) learn to measure are important. In particular, a match in values ​​and culture, teamwork, communication skills, a sense of humor, etc. are important.»

EVO agrees with this and adds a practical emphasis: the final decision remains with the people «especially when it comes to soft skills and cultural match.» This is not just a beautiful phrase — there is a real assessment problem behind it. The algorithm can check the keywords in the resume, but it is not able to understand whether the person fits into the team, whether he has the potential for growth that is not visible from the formal lines of experience.

There is also a purely technical problem, which Railsware describes: the diversity of job titles and the experience behind them. A recruiter understands that «Technical Lead» in one company may correspond to «Senior Engineer» in another. An algorithm without additional context does not. This is why Levi9 uses AI as a prioritization tool, not a selection tool: it is useful «when it comes to vacancies with a large number of applications» — that is, where it is physically difficult for a person to review all the resumes, but the final word remains with the recruiter.

Transparency: Do candidates know that they are being evaluated by AI?

The issue of transparency is a separate topic that does not yet have a single standard in the market. Some companies clearly communicate the use of AI tools to candidates, others do not.

Railsware and N-iX explicitly state that they require prior notice and consent before using AI notetaker. EVO notes that a candidate can opt out of the recording, but these are rare cases.

This raises an important question for the market: is it enough to inform about the notetaker? Should the candidate know that his resume is being analyzed by an algorithm that forms a match score, that AI generated questions for his interview? As long as there is no legislative regulation in Ukraine, each company makes these decisions independently.

Trends that are forming right now

Having collected all the answers, we can identify several clear trends that will determine hiring in the coming years.

From point tools to integrated ATS. Most companies started with stand-alone AI tools — Copilot here, ChatGPT there. But now SoftServe and Ciklum are describing the integration of AI directly into ATS systems. This means that automation becomes systemic, not situational.

Standardization through AI. Several companies mention standardization as one of the key values ​​of AI tools: uniform evaluation criteria, consistent feedback, unified candidate profiles. This reduces the human factor — but at the same time can reduce flexibility.

AI for understanding the vacancy, not just finding candidates. A separate trend is the use of AI at the stage of role analysis and candidate profile formation. Here, AI acts not as a search engine, but as an analytical partner for the recruiter.

Human as a verifier. Regardless of the level of automation, in none of the companies surveyed does AI make final decisions on its own. This may sound obvious — but it is an important signal to candidates: there is always a human behind your offer or rejection.

A few tips for candidates

If you are currently looking for a job in IT, here are the practical conclusions from all of the above.

1. Your resume is likely the first thing an algorithm reads. In companies with a high application flow, AI is helping to prioritize candidates. This means that keywords, a relevant structure, and a clear statement of experience are more important than ever. But that doesn’t mean you should stuff your resume with keywords—algorithms are already smart enough to distinguish relevant experience from spam.
2. You will most likely be recorded during the interview. AI notetaker has become the standard. If this detail is important to you, you have the right to ask or refuse. But understanding that notes can be reviewed later is useful for the quality of answers.
3. The final decision always rests with the human. At least in the companies that participated in this article. This means that «not fitting the algorithm» is not a sentence. A human can see what the screener missed.
4. Soft skills and cultural fit are not automated. These are the things companies say are the hardest to assess—and the most important. That’s good news for candidates: what makes you unique is still beyond the reach of AI.

«Artificial intelligence makes a sign for each government decision.» Bornyakov told how AI helps government officials predict who benefits from certain bills

On the topic

«Artificial intelligence makes a sign for each government decision.» Bornyakov told how AI helps government officials predict who benefits from certain bills

Deepfake, perfect answers, simulated network failures and seven other AI tricks of candidates that recruiters recognize in a minute

On the topic

Deepfake, perfect answers, simulated network failures and seven other AI tricks of candidates that recruiters recognize in a minute

Read the country’s main IT news in our Telegram

On the topic

Read the country’s main IT news in our Telegram