Search for candidates in plain English, not Boolean
Boolean search asks you to guess the words on someone's profile. Natural-language search asks you to describe the person you want. The second is a better description of what you actually know at the start of a search, and it does not degrade when a candidate happens to call the same job something else.
You write a paragraph. You get people, in order, with reasons.
You need a Nyne account to run a search. Talent Search spends your Nyne workspace credits, so it signs you in first.
Already signed in? Open Talent Search. No Nyne account yet? Create one at nyne.ai.
What a brief looks like
A good brief reads like an explanation to a colleague rather than a query. It says what the person should have built or owned, at what scale, and what would make you pass. For example:
"Senior engineers who have owned production machine learning or LLM systems end to end: evaluation harnesses, fine-tuning and feedback loops, inference orchestration, and measurable accuracy or latency improvements. Strong Python and system design, with real MLOps experience. Prioritise high-ownership builders."
That brief is doing several things a Boolean string cannot. It states seniority as a level of responsibility rather than a title, names outcomes instead of technologies, and expresses a preference that is used for ranking rather than a filter that silently deletes people.
Filters are still there when you need them
Plain English is the default, not a restriction. Some constraints are genuinely binary - a work location, a language, a specific set of titles - and those belong in a filter where they behave predictably. Twenty structured fields are available alongside the brief, and every one of them is optional.
- Hard constraints
- Locations, titles, companies, universities, degrees, fields of study, industries, keywords and languages.
- Employer constraints
- Headcount range, total funding raised, funding stage, named investors and last funding date.
- Preferences
- Anything you would rather express as "prioritise" than as a rule belongs in the brief, where it shapes the ranking instead of removing people.
Why the evidence matters more than the ranking
Any system can return a sorted list. The question that decides whether you trust it is what happens when it is wrong. Each candidate here carries a short summary of why they matched and a set of evidence items naming the phrases from their history that triggered each criterion.
This makes the failure mode visible. A candidate who scored well on "evaluation harnesses" because they wrote the phrase once in an unrelated context looks obviously wrong when you can see the phrase. Without the evidence, that person just looks like a four out of five.
Signing in
Nyne Talent Search is an account product. Searching runs against your Nyne workspace and spends your workspace credits, so you sign in first.
Common questions
Is natural language search better than Boolean?
For describing a role, generally yes: it captures seniority, scope and outcomes that no keyword string expresses well, and it does not miss candidates who use different vocabulary. For hard constraints such as a location or a specific title list, a structured filter is more predictable. Nyne Talent Search offers both together rather than making you choose.
Do I need to learn a query syntax?
No. There are no operators, no quoting rules and no field prefixes to learn. You write the brief as a paragraph of ordinary English, and any hard constraints go into labelled filter fields.
How do I know why a candidate was matched?
Every candidate shows a written summary of the match plus individual evidence items, each naming the phrase from that person's history that satisfied a criterion. Opening a candidate shows the full set alongside their work history and education.
Can I still search by job title and location?
Yes. Titles, locations, companies, universities, degrees, industries, keywords and languages are all available as structured filters, and they can be combined with a written brief in the same search.