AI that finds candidates.
Not AI that just filters them.

Keyword matching is ctrl+F applied to a database. It finds candidates who used the exact words in your job description. Vector matching understands context — synonyms, related concepts, implied skills. It also understands what's missing. A CV heavy on technical depth but light on people management will match a senior IC role better than a VP role, even if neither brief mentions 'individual contributor'. That nuance is invisible to keyword systems.

Yes, it can — if built carelessly. AI trained on historical hiring data inherits historical biases. We don't train Yena's matching on your historical hires precisely to avoid this. The vector model is trained on language understanding, not on 'people who got hired here before'. We also surface the matching rationale so you can check what the AI weighted, and override it. The EU AI Act classifies AI recruitment tools as 'high risk' — which means transparency and human oversight aren't optional. We built for that from day one.

Article 22 gives individuals the right not to be subject to solely automated decisions that significantly affect them — and the right to an explanation when they are. Yena provides exportable rationale for every AI ranking. Human recruiters make the final shortlisting decision. The AI ranks and recommends; it doesn't reject. You can configure candidate-facing communications to explain that human review occurs. Data is EU-hosted, consent workflows are built in, and retention policies are configurable per jurisdiction.

Professional and technical roles where career history is meaningful — executive search, engineering leadership, commercial directors, finance, legal, specialist technical. It's less useful for high-volume early-career hiring where volume processing is the challenge, not semantic matching. If you're screening 800 barista applications, keyword filtering plus a structured test is probably more efficient than vector matching. If you're finding a CFO for a Series C FinTech, vector matching is genuinely different.

No minimum database size. Vector matching works on a database of 10 or 10,000. Smaller databases just return fewer results — the quality of matching doesn't degrade. Where it compounds in value is for retained search firms and in-house teams with years of candidate history. Those 3,000 candidates who applied 18 months ago? Yena can re-rank them against a new brief in under a second. Most teams find their first AI-surfaced hire from their existing data within the first two weeks.