Over the past year, we’ve experimented extensively with our hiring process at Notix.Games. We have conducted dozens and dozens of interviews, spoken with a large number of candidates, made several great hires, gathered feedback from both candidates and hiring managers, and rethought our familiar approaches.
As a result, we have shortened our hiring process, automated a significant portion of our sourcing, strengthened background checks, and introduced new tools for working with interviews. At the same time, some core principles on which we based our recruitment process a few years ago have remained unchanged.
Why did we decide to change our process? Because the context in which recruiting operates has shifted.
Today, both candidates and companies use Large Language Model (LLM) tools to prepare for interviews, complete test assignments, find information, and automate routine tasks. The ability to work effectively with these tools is becoming a core professional competency—and we view it as a major advantage.
For us, the question about AI in hiring isn’t whether to use it, but how to evaluate a candidate’s skills under these new conditions. We want to understand how someone applies modern tools, how deeply they know their domain, and whether they can critically evaluate the output, verify its accuracy, and take ownership of the final decision.
Meanwhile, both sides of the hiring table have found themselves in a new reality: candidates have more ways to showcase their knowledge and skills, while companies have additional tools to source, evaluate, and verify information. To keep this interaction productive, we need to establish shared, transparent rules for working with AI.
Over the past year, we have rebuilt specific parts of our process to adapt to these new conditions. In this article, we’ll share what we’ve changed, what we’ve kept the same, and why.
1. Fewer Interview Stages, More Substance
When we first built the recruiting process at Notix.Games, it was important for us not only to assess a candidate's hard skills, but also to understand how well they fit our culture and how they would collaborate with the team.
To do this, we used to have a separate HR interview stage. Over time, we realised that many of the questions covered during that call were already coming up in other meetings. We were repeatedly asking about motivation, teamwork experience, autonomy, and problem-solving approaches—even though we could evaluate these qualities throughout the entire process.
So, we removed the standalone HR interview as a required stage.
However, we didn’t drop culture fit assessment. On the contrary, it is now embedded into every stage where it naturally belongs.
During a technical interview, we can evaluate not only the accuracy of a solution, but also how someone navigates ambiguity, asks questions, receives feedback, and explains their reasoning. When discussing past experience, we can gauge the level of ownership the candidate actually had, how they collaborated with peers, and what they consider to be their own tangible results.
Additionally, this is where the combination of Metaview and LLMs helps us: after the interview, the transcript or summary is scored against our competency framework, the same one we use competency-based interviews. This doesn't replace human evaluation, but it effectively highlights key markers and nuances that the interviewer or hiring manager should pay attention to.
This allows us to see the person much more clearly in a professional context, rather than judging them solely by their answers to pre-scripted questions.
For example, if a candidate shares that they independently improved a complex workflow, we want to understand more than just the final outcome: we look at how they spotted the problem, what alternatives they weighed, who they aligned the changes with, and how they measured the impact of their work.
That tells us far more than just asking a standalone question about whether someone considers themselves proactive.
- What changed: The process is shorter, cultural evaluation is no longer confined to a single meeting, and LLM-driven competency analysis of interviews helps better surface important behavioural signals.
- Why we made this decision: Every stage should bring new insights and help drive a decision. If the same goal can be achieved within existing touchpoints, a separate stage isn’t necessary.
2. An Interview Is No Longer Enough: Why We Rely More on Reference Checks
Talking to a candidate remains a vital part of hiring, but we are less and less inclined to view that as the sole source of truth.
An interview is time-limited. The candidate talks about their experience, answers questions, and solves problems, while the interviewer tries to piece together a picture of their professional level. Yet even a great conversation doesn’t always allow you to verify every critical detail.
With the rise of LLMs, candidates’ capabilities have expanded. Modern tools help structure thoughts, find solutions, analyse information, and prepare for complex technical questions. This shifts the very nature of the interview: today, it’s important to evaluate not just the final answer, but how the candidate arrived at it, what tools they used, and how thoroughly they understand the output.
That’s why we’ve strengthened our post-interview background checks and become much more thorough with reference checks.
Our goal isn’t merely to confirm employment dates or job titles at a specific company. We want to gain a deeper understanding of a person’s actual contribution to projects, their level of autonomy, the nature of the challenges they tackled, and how they interacted with the team.
This additional context helps us cross-reference different parts of a candidate's career story. For instance, if someone talks about launching a major project, we want to understand which part of the outcome they owned personally, what decisions they made, and what specifically changed as a result of their work.
This doesn’t mean we suspect every candidate or try to hunt down discrepancies at all costs. Rather, we want to make hiring decisions based on a more complete picture.
- What changed: The interview is now treated as one part of a broader evaluation, complemented by professional references and background verification.
- Why we made this decision: It’s important for us to assess not just self-presentation skills, but real-world professional experience, personal impact, and potential.
3. Candidates Using AI in Interviews: We Encourage Open Use
We don’t view LLM use as inherently negative. On the contrary, knowing how to work effectively with modern tools can be a significant professional advantage.
For example, using AI during a take-home assignment can be entirely appropriate. In day-to-day work, a professional often doesn't need to perform every single operation manually; instead, they need to frame the problem, choose the right tools, verify the output, and take accountability for the result.
In this context, the ability to use LLMs thoughtfully points to a mature professional approach.
However, for an accurate evaluation, we need to understand how the candidate uses these tools and what they can achieve with them.
Imagine a candidate completes a test assignment using an LLM. That fact alone doesn’t make the result any less valuable. What interests us far more is whether the candidate understands the proposed solution, can walk us through the reasoning, verify the accuracy of the output, spot flaws, and justify their own choices.
If the rules of an assignment allow AI, we see no reason to pretend these tools don’t exist. On the contrary, transparent usage lets us evaluate the exact skills that actually matter on the job.
At the same time, the evaluation conditions must match the goal of the specific stage. If we want to see how someone solves a problem using modern tools, we want to observe that workflow openly. Conversely, if the goal of a stage is to test unaided knowledge or core fundamentals, that expectation must be communicated to the candidate upfront.
Our stance is simple: we welcome AI where it makes sense, and we consider transparency around its use to be an essential part of professional communication.
Because of this, we have also introduced a set of tools to help detect potential cases of undisclosed LLM use during live interviews.
An important caveat here: we do not treat technical signals as absolute proof of dishonesty. Such tools can be mistaken, and a single flag rarely explains a candidate’s behaviour with 100% certainty. Their role is simply to highlight a potential issue that needs to be weighed within the broader context of the conversation.
We have no interest in starting a technological arms race between candidates and interviewers. Our goal is to build a process where both sides understand what is being evaluated, and where modern tools help paint a more accurate picture of competencies.
- What changed: We actively use tools that help verify evaluation integrity, and we draw a clearer line between open AI usage and attempts to conceal it when unaided skill is being tested.
- Why we made this decision: What matters to us isn't how well a candidate can get by without modern tools altogether, but how they apply them in practice, how they think, make decisions, and own the outcome.
4. From Manual Search to Our Own AI Sourcing Platform
Technology hasn't just changed interviews—it has transformed the recruiter’s workflow long before the first conversation with a candidate takes place.
Finding relevant professionals requires combing through numerous external sources, analysing career histories, and re-engaging candidates we’ve spoken with in the past. Much of this work is repetitive and eats up time that would be better spent on more meaningful tasks.
To solve this, we built our own internal AI sourcing platform.
It leverages DataForSEO, Crustdata, and the OpenAI API to automate a large portion of search across external platforms and our internal candidate database.
Instead of relying solely on manual searches, we can use our own infrastructure to surface relevant profiles and work much more effectively with the data we’ve already accumulated.
For us, it was important not just to plug in an off-the-shelf AI tool, but to build a custom solution tailored to our specific needs and workflows.
That said, automation doesn’t replace the need to understand the market. Finding a profile with the right set of keywords is not the same as finding the right person for the role. The context of their experience, the scale of their past challenges, their motivation, and their potential impact all require human judgement.
We want technology to take over the repetitive heavy lifting so recruiters can focus more on search quality, candidate relationships, and deeply understanding professional backgrounds.
- What changed: Our proprietary AI platform automates a substantial share of sourcing across external channels and our internal database.
- Why we made this decision: A recruiter’s value doesn't lie in the number of profiles they manually review, but in their ability to find the right people, understand their background, and help the business make the right hiring call.
5. Interview Intelligence with Metaview: Context Instead of Memory
Another area of change is how we handle information gathered during interviews.
When multiple team members evaluate the same candidate across different stages, preserving the context of those conversations and being able to revisit specific details is crucial. At the same time, an interviewer can’t always listen attentively, ask follow-up questions, and jot down every nuance simultaneously.
To address this, we use Metaview—an AI-powered interview recording and documentation tool that lets us revisit conversations whenever needed.
This helps us reconstruct discussions accurately, verify details, and base our post-interview debriefs on actual facts rather than rough impressions.
For instance, if a candidate discusses their contribution to a specific project across multiple stages, it’s helpful to go back to the initial discussion and compare the details accurately. Or, if a hiring manager wants to clarify a specific aspect of a candidate's background, they don’t always need to schedule another long call: sometimes, reviewing the notes and transcript from a previous interview is enough.
We don’t view recording as a substitute for an engaged conversation. It is a supplementary tool that helps us work with information more systematically.
And, of course, using tools like this always requires participant consent, strict confidentiality, and compliance with data privacy standards.
- What changed: We can now capture, store, and revisit interview insights using a dedicated tool.
- Why we made this decision: High-stakes hiring decisions shouldn't depend solely on what an interviewer happens to remember after a call. The better we preserve context, the more objective and focused our candidate discussions become.
6. Networking and Employee Referrals: A Core Part of Our Strategy
One of the changes we are most proud of has nothing to do with technology—it’s about professional relationships.
Throughout the past year, we have consistently nurtured our network, stayed in touch with industry peers, and encouraged our team members to tap into their own professional circles to refer candidates.
Today, we hire predominantly through our network, and we’ve seen a clear positive impact on hiring quality.
Why does this work?
First, professional networks help us reach people who aren’t actively job hunting and rarely show up in standard sourcing channels.
Second, a connection often starts long before a relevant opening even exists. We get a better sense of a person’s background, professional interests, and career goals over time, while the candidate gets to know our company beyond a standard job description.
Third, peer referrals add valuable context. If a team member knows someone from a past project or a professional community, they can highlight that person's core strengths and the types of challenges where they truly shine.
For example, an employee might refer a former colleague with whom they tackled complex engineering problems. That recommendation helps us know which parts of their experience to explore and what questions to ask during the interview. However, the final hiring decision still comes down to how well the candidate’s skills match the requirements of the specific role.
There is an important nuance here: a referral helps open the door, but it is never an automatic pass.
We hold every candidate to the exact same evaluation standards—whether they came through our professional network, applied via a job board, or were found through AI sourcing.
Networking doesn’t replace recruiting. It makes it more meaningful and helps us connect with people where there is a strong mutual fit.
For us, this is also proof that advancing technology doesn’t diminish the value of human connection. On the contrary, when automation handles the technical heavy lifting more efficiently, it frees up time to cultivate relationships that can never be built with a single search query.
- What changed: Our network has become one of our primary hiring channels, and nurturing it is now a deliberate, consistent part of our recruiting strategy.
- Why we made this decision: Long-term professional relationships help us find relevant people and understand their track record before the formal hiring process even begins.
7. What We Kept Unchanged
Allthese changes might make it look as though we have completely overhauled our hiring philosophy. But that’s not the case.
We still believe that recruiting should be transparent, respectful, and value-driven.
In our very first article about how we hire at Notix.Games, we talked about trust, ownership, a results-oriented mindset, open communication, and respect for candidates' time. Those principles remain just as relevant today.
We still evaluate outcomes, not just CVs. We care about what a person has actually accomplished, what decisions they've made, and what value they've created. Knowing how to leverage modern tools is simply part of delivering those results today.
We still value honesty. We aren’t looking for a candidate who claims to know the answer to every question. We are far more interested in someone who can speak openly about their strengths, their limitations, and their willingness to learn.
We still care deeply about a culture fit. We just evaluate it organically throughout the entire process rather than isolating it in a separate stage.
We still strive to respect candidates' time. A leaner process cuts out redundant meetings and keeps the focus on interactions that genuinely drive the final decision.
We still view hiring as a two-way street. The company evaluates the candidate, and the candidate evaluates the company. They need a clear picture of upcoming challenges, team dynamics, and working conditions to make an informed choice of their own.
And finally, we still believe that technology never absolves us of responsibility for the final hiring decision. Automation can help surface a profile, structure notes, and flag additional signals. But it cannot replace an understanding of professional context, human communication, and sound judgement.
We don’t want to cling to an old process just because it feels familiar. Nor do we want to change our approach simply for the sake of adopting shiny new tools.
What matters to us is keeping the elements that help us make high-quality decisions and rethinking the ones that no longer add enough value.
8. Recruiting as Continuous Adaptation
If the past year has taught us anything, it’s that effective recruiting isn’t a static blueprint you can set once and repeat forever.
The market shifts. Candidate expectations and business needs evolve. New tools emerge, opening up fresh possibilities while raising new questions. Because of this, the hiring process itself must be regularly revisited.
We continue to gather feedback, analyse our hiring outcomes, discuss the process with hiring managers and candidates, and test which adjustments actually work in practice.
We don’t aim to automate everything indiscriminately, nor do we believe every stage should be preserved simply because "that’s how we’ve always done it." Our goal is to understand what helps us find the right people, assess their capabilities more accurately, and build teams that deliver real results.
Technology helps us cast a wider net, process information more efficiently, and dedicate more energy to meaningful evaluation. Professional networks help us connect with the right talent and stay close to the market. And our core values provide the compass that lets us make decisions not just quickly, but thoughtfully.
To us, the evolution of recruitment isn’t a choice between automation and a human touch. It’s about finding the right balance of technology, domain expertise, and trust.
And that is the exact balance we continue to build at Notix.Games.




.jpg&w=3840&q=100)
.png&w=3840&q=100)