AI-polished job applications hide how people really learn. Dmitry Zaytsev talks through hiring using simple games to reveal learning readiness early. Dmitry’s case study shows why behavioural signals like persistence and attention matter more than credentials for talent, L&D, onboarding, and building learning cultures that actually stick in modern organisations.
In many organisations, learning and development begins only after a person has signed a contract. By that point, L&D teams inherit a challenge they did not create: they are asked to build learning cultures, yet a noticeable share of new joiners arrive without basic learning habits. The CV and the interview process rarely reveal this, especially in an era where most application documents are polished by AI.
Instead of evaluating candidates through documents, we treated one hiring cycle as a behavioural experiment
Over the past year I have been exploring a different approach. Instead of evaluating candidates through documents, we treated one hiring cycle as a behavioural experiment. For a junior business development role, we removed CVs from the early stages and asked candidates to complete four short online games, a self-pitch task and a simple sales simulator. Across two weeks the funnel looked like this:
- Around 500 people responded to the vacancy
- 165 completed registration
- 99 started the tasks
- 46 finished all four games
- 34 entered the simulator
- 24 achieved at least one confirmed sale
- 3 were hired
Experiment results
The tasks themselves were intentionally simple; they didn’t require knowledge of business development or prior experience. What they did require was curiosity, attention, persistence and the ability to follow instructions. The sales simulator added a layer of social logic, basic argumentation and situational adaptation.
What stood out was not the performance of the strongest candidates, but the consistency of the patterns we saw among those who disengaged. Most dropouts happened not because the tasks were difficult but because people stopped reading instructions, left tasks unfinished or did the minimum required to progress. In other words, they showed the same behaviours that later undermine real workplace learning.
Equally striking was the behaviour of those who progressed furthest. This smaller group tended to complete every step without reminder, adapted quickly when the simulator changed context and often did a little more than required. They demonstrated the same traits that help employees succeed in learning environments: initiative, self-management, tolerance of ambiguity and constructive response to challenge.
Because the tasks were time limited and sequential, we could observe early patterns that normally emerge months after onboarding. For example, candidates with stronger attention and accuracy scores, and those who answered in their own words rather than relying heavily on AI tools, tended to perform better in the simulator and in the final interviews. Those with higher social intelligence scores adjusted their communication approach, increased conversion rates and recovered better from mistakes.
Actions for L&D
From the perspective of L and D, several lessons became clear.
- Learning readiness is visible long before a learner reaches a classroom
Behavioural cues that matter for learning, such as persistence, attention, follow through, appear within minutes of giving someone a new task. These signals are far more reliable than CVs or self-descriptions. - Small game-like challenges surface habits that traditional assessments miss
A short interactive task reveals how a person deals with uncertainty, instruction changes or light pressure. These are core learning behaviours that no multiple-choice test can capture. - Learning and hiring are connected more tightly than most organisations assume
If early tasks expose weak self-management, it is likely to resurface during onboarding and training. Conversely, candidates who consistently complete tasks, ask clarifying questions and adapt quickly often become effective learners regardless of previous experience. - AI makes formal signals less trustworthy and behavioural signals more valuable
When large language models can produce polished CVs at scale, organisations need alternative ways to observe how people actually think and behave when facing new challenges.
For L&D practitioners, this suggests a more integrated approach. Rather than waiting for employees to fail, organisations can work with people teams to design early micro challenges that reveal learning habits and begin development before formal training occurs. These challenges do not need to be complex. What matters is that they generate observable behaviour: how someone approaches a first step, responds to feedback, resets after an error or persists when the task becomes slightly uncomfortable.
Challenge and response
The broader lesson is simple. Learning does not start in the classroom. It starts the moment a person encounters something unfamiliar. The earlier organisations notice how people behave in those moments, the better prepared they are to build learning cultures that work in practice rather than in theory.
Dmitry Zaytsev is the Founder of Dandelion Civilization
