Natural Language Candidate Search: How a Plain-English Brief Replaces the Boolean String

By Sparsh Goyal, Founder at The Cognitive. Published 2026-10-04 · Last updated 2026-10-05

• Natural language candidate search turns one plain-English sentence into search filters and a ranked list of candidates, with no Boolean operators.
• The Cognitive reads the brief into filters you can edit, such as "Remote EU" read as Germany plus 7 other countries, and anyone who misses a must-have drops below the people who meet them all.
• Use it for a fast ranked list, and keep Boolean for exact exclusions, free Google X-ray searches and searches you must repeat word for word.

Natural language candidate search lets you describe the person you want in one plain-English sentence, and the tool turns it into search filters and a ranked list. No AND, OR or brackets. The Cognitive does this across ~900M public profiles: it reads your brief into filters you can edit, then ranks everyone against your must-haves.

Recruiters need the shortcut. Ashby's data shows about 291 applications per hire in 2026, up from about 100 in early 2021, and Gem's 2026 benchmarks say direct sourcing delivers 11% of hires from just 2.6% of applications. Candidate sourcing works. Writing the Boolean string is the slow part, and that is the part natural language replaces.

Full disclosure: I run The Cognitive. Everything I say about the other 6 tools comes from their own websites, read on 4 October 2026.

What is natural language candidate search?

Natural language candidate search is a way to find candidates by typing a sentence a hiring manager would say, such as "senior backend engineer, Go, payments, remote in Europe". The software works out the job title, seniority, skills and location, searches a profile database, and returns people ranked by how well they fit. You will also see it called AI candidate search, AI talent search, semantic search or AI people search.

Boolean search is the older method. You write the logic yourself: ("backend engineer" OR "backend developer") AND (Go OR Golang) AND payments. The engine matches those exact words and nothing else. Our guide to Boolean search operators covers the logic, the free Boolean search generator writes a string from a job title, and 51 Boolean strings by role gives you ones to copy.

The difference comes down to who does the translating. With Boolean, you turn the job into keywords. With natural language, the tool does it, and your job moves to checking its reading.

How does a plain-English brief become search filters?

The tool reads your sentence the way a sourcer reads a job intake note. It pulls out each requirement and maps it to a field the database can search: title, seniority, function, location, skills, years of experience.

Here is a real one. On 30 September 2026 I typed this brief into The Cognitive:

Senior backend engineer with 5+ years of Go who has built payments systems, Remote EU

It came back as editable filters: Job title "Backend Engineer +1", Seniority "Senior", Job function "Software", Location "Germany +7" and Skills "Go +1". The interesting one is the location. "Remote EU" is not a place a database can search, so the tool read it as Germany plus 7 other countries. A Boolean user would have had to list those countries by hand.

That reading is a guess, and you should check it. If your company can only hire in 3 of those countries, open the location filter and fix it. While the search runs, The Cognitive shows 2 steps: "Reading your brief and finding relevant candidates", then "Ranking candidates based on your requirements". It ends with "Search complete" and the number of candidates found.

Good tools make the reading visible. Juicebox says on its PeopleGPT page that it "understands your searches, configures filters". If a tool hides the filters it built, you cannot tell a bad brief from a bad search.

How does AI candidate matching rank people against your must-haves?

Filters decide who gets searched. Ranking decides who you see first. This is where AI candidate matching earns its name, and where tools differ most.

In The Cognitive, every person the search returns is judged against each must-have in your brief. Anyone who misses one drops below the people who meet them all. Nobody is removed. So the top of the list is people who meet every must-have, and further down you still see the near misses, which matters when the perfect profile does not exist.

3 rules keep that ranking honest:

• Proof comes from the profile. A required licence, degree or language only counts as met when the profile shows it.
• Every card says why. Each result opens with a "Why this match" line of 1 or 2 reasons traced to the profile, such as "Current role: ..." or "Go listed in skills". You get a reason you can check, not a number.
• Thin pools get widened in the open. When too few people qualify, the search relaxes its softest filters and tells you what it broadened.

Your own decisions feed back in. Once you have made 8 or more decisions, including 3 or more shortlists, what you shortlist and pass on shapes later ranking. Everyone found stays with the role, and later searches skip people you have already seen.

Other tools score instead. Gem says each result comes with "a match score with clear reasoning", and Pin says its AI "scores every candidate against your job requirements". A score is quick to sort by. A stated reason is quicker to check.

Type the brief you would have written as a string One sentence, read into editable filters and ranked against your must-haves across ~900M public profiles. Start free

Where does natural language search beat Boolean, and where does it lose?

Neither method wins everywhere. Most good sourcers will end up using both, for different jobs.

Where natural language wins

• Speed to a first list. One sentence replaces a string you would otherwise build, test and fix over several rounds.
• Synonyms you forgot. A Boolean string misses "Golang" if you only wrote "Go". A tool that reads meaning catches both. hireEZ describes its Deep Search as semantic search that "broadens your pool of sourced candidates".
• Ranking. Boolean returns a flat set of matches. Natural language tools sort by fit, so the best people are on page 1.
• Hiring managers can use it. A founder who has never seen a bracket can type a brief and get a usable list.
• Fuzzy requirements. "Has built payments systems" is hard to express as keywords. A model reading the whole profile has a better chance.

Where Boolean still wins

• Exact control. A string does what it says. If you must exclude a company, a title or a term, NOT is precise.
• Repeatable and auditable. The same string returns the same logic tomorrow. A model's reading of a sentence can shift.
• It works anywhere. With small tweaks, the same string runs on LinkedIn, Google, Indeed and Dice. A natural language tool only searches its own database. See our LinkedIn X-ray guide for the Google route.
• It is free. A Google X-ray costs nothing. Every tool on this page charges for full use.
• Niche jargon. A rare certification or an internal product name may be misread by a model. A quoted exact phrase will not.

My rule: start with natural language to get a ranked list fast, then reach for Boolean when you need an exclusion, a source the tool does not cover, or a search you must repeat word for word.

Which recruiting tools offer natural language candidate search?

At least 7 well-known tools now take a sentence instead of a string. I read every cell below on each vendor's own site on 4 October 2026. Profile counts are their own figures, and I give the pricing model only.

1. The Cognitive: best for seeing why each person ranks where they do

The Cognitive is AI recruiting software that sources candidates across ~900M public profiles and runs a live AI video interview with the people you shortlist. The search reads your brief into editable filters and ranks everyone against your must-haves, with a "Why this match" line on each card.

The trade-offs: it does not search your own ATS history, it has no Chrome extension, and email and SMS outreach needs the Sourcing Pro plan. There is no LinkedIn InMail. The AI sourcing tool page walks through the full search.

Pricing model: published monthly plans. AI Sourcing starts at $49/month and AI Interview at $99/month. Credits: 1 per candidate a search returns, 5 to reveal an email, 10 to reveal a phone number, charged only when a value comes back. There is a free trial with 100 sourcing credits.

2. Juicebox: best for natural language search on a free plan

Juicebox's PeopleGPT is the product most recruiters picture when they hear the term. It says it "configures filters, and supports full natural-language search" across "800M+ profiles from 30+ data sources", and that it "evaluates up to 5,000 profiles" to find the best match. The free plan makes it the easiest to try. Our Juicebox review covers it in depth.

Pricing model: free plan; Starter and Growth are priced per seat per month with an annual discount; Business is custom.

3. SeekOut: best for enterprise teams that want search tied to their ATS

SeekOut's homepage invites you to "describe the role in your own words", then says it searches 1B+ profiles, ranks the strongest matches and drafts outreach. It is built for enterprise teams; the entry plan is self-serve, and the tiers that connect your ATS are custom. More in our SeekOut review.

Pricing model: Sourcing Core shows a price, monthly or annual, with a 14-day free trial and no card; other tiers are custom and billed annually.

4. hireEZ: best for AI agents on top of your ATS

hireEZ asks you to brief its agent the way you would brief a sourcer, and says it explains the reason behind every match. It searches 45+ platforms, sits on top of your ATS, and adds voice AI screening. Deep Search widens a pool that strict keywords made too small.

Pricing model: a solo Growth plan shows a monthly price with a 7-day trial; Enterprise is quoted by sales, priced "to the stack hireEZ replaces rather than a flat per-seat rate".

5. Gem: best for searching your ATS and CRM in the same query

Gem's AI Sourcing Agent turns a plain-language description into a tailored search. Its edge is scope: it "searches 800M+ profiles, past candidates in your ATS and CRM, and ideal profiles". If your best people already applied once, that matters, and our guide to candidate rediscovery shows how to work that pool. See our Gem review for the rest of the product.

Pricing model: custom pricing based on company headcount (FTE), through sales; its startup program is free for 6 months under 30 employees, with reduced rates from 30 to 100.

6. Pin: best for pasting a job description as the search

Pin's quickest route skips the brief: "Paste a job description and get a ranked candidate list in seconds." It searches 850M+ profiles and scores each candidate against the job's requirements. The trade-off is that a job description is a wish list, so you get its padding along with its must-haves. Our Pin alternatives guide compares it with similar sourcing tools.

Pricing model: free plan; Solo and Professional are priced per user and billed annually; Business is on request.

7. LinkedIn Hiring Assistant: best for teams already paying for LinkedIn Recruiter

Hiring Assistant works the other way round: it asks you questions about the role to build a sourcing strategy, then runs "dozens of searches across LinkedIn". It learns from your past Recruiter activity on similar roles. It only searches LinkedIn, and LinkedIn says Hiring Assistant 2 rolls out in early November. Our LinkedIn Hiring Assistant review asks whether it is worth the add-on.

Pricing model: an add-on to LinkedIn Recruiter, sold through sales; no price on the website.

For the full ranking of sourcing tools, including ones without natural language search, see best AI sourcing tools.

How do you write a good brief for natural language search?

A brief is a Boolean string without the punctuation. The same discipline applies: say what is required, what is preferred, and where. 5 habits help:

• Lead with the title a candidate would use. "Backend engineer" beats "Platform Rockstar".
• Put numbers on experience. "5+ years of Go" is checkable. "Experienced" is not.
• Name the proof. "Has built payments systems" gives the tool something to find in a profile.
• Be exact about location. "Remote EU" works, but check which countries it becomes.
• Keep it to 3 or 4 must-haves. Every extra must-have pushes more good people down the list.

Here are 5 briefs by role, written to that pattern. Copy one, change the details, and check the filters it produces before you trust the list.

1. Software engineer: "Senior backend engineer with 5+ years of Go who has built payments systems, Remote EU"
2. Nurse: "Registered nurse with an active RN licence and 3+ years in ICU, Texas, open to night shifts"
3. Sales: "Enterprise account executive with 4+ years selling SaaS to banks, based in New York or Boston"
4. Design: "Senior product designer with 6+ years in B2B SaaS who has shipped a design system, London or remote UK"
5. Customer support: "Customer support team lead fluent in Spanish and English who has managed 5+ agents, Mexico City"

In the nurse and support briefs, the licence and the language are must-haves a tool should only count when the profile shows them. In The Cognitive, a profile that does not show the RN licence drops below the ones that do.

When a brief returns too few people, loosen the softest requirement first, usually the years or the city, before you touch the skill that defines the job.

How do you start with natural language candidate search?

Take your hardest open role, the one where your Boolean string returns either 4 people or 4,000. Write it as one sentence with 3 or 4 must-haves. Run it, open the filters, and fix anything the tool misread. Then read the top 20 cards and ask whether each reason holds up against the profile.

If you want to run that test on The Cognitive, sign up and type the brief. There is a free trial with 100 sourcing credits.

Run your hardest role as one sentence See who meets every must-have across ~900M public profiles, and why, before you write a single string. Start free

Sources

• Ashby, Recruiter Productivity Trends Report (applications per hire), read 4 October 2026: ashbyhq.com
• Gem, Key takeaways from the 2026 Recruiting Benchmarks Report, read 4 October 2026: gem.com
• Juicebox, PeopleGPT page, read 4 October 2026: juicebox.ai/peoplegpt
• Juicebox, pricing, read 4 October 2026: juicebox.ai/pricing
• SeekOut, homepage, read 4 October 2026: seekout.com
• SeekOut, pricing, read 4 October 2026: seekout.com/pricing
• hireEZ, homepage, read 4 October 2026: hireez.com
• hireEZ, Deep Search, read 4 October 2026: hireez.com/deep-search
• hireEZ, pricing, read 4 October 2026: hireez.com/pricing
• Gem, AI Sourcing Agent, read 4 October 2026: gem.com/product/ai-sourcing
• Gem, pricing, read 4 October 2026: gem.com/pricing
• Pin, homepage, read 4 October 2026: pin.com
• Pin, pricing, read 4 October 2026: pin.com/pricing
• LinkedIn, Hiring Assistant, read 4 October 2026: business.linkedin.com

All posts · thecognitive.io