AI Candidate Sourcing: How CloudHire Finds Talent Across 700M+ Profiles

What is CloudHire's AI candidate sourcing?

CloudHire's AI candidate sourcing is a job-anchored search engine that lets businesses find candidates across an index of more than 700 million professional profiles using four modes: plain English, a pasted job description, boolean logic, or manual filters. Every search is tied to a specific job, so results arrive pre-ranked with an ATS match score on each candidate, verified Cloud-ID profiles boosted in the list, and contact details ready to unlock. Recruiters can even run a real search before creating an account, with names redacted and contacts blurred.

That single design decision, anchoring search to a job rather than to a keyword string, is what separates CloudHire from a classic resume database. The system does not just retrieve profiles that mention "React" or "Sales Manager"; it evaluates each profile against the role you are hiring for and tells you why a candidate fits or does not. This guide walks through every sourcing capability in detail. For the full library of platform guides, see the For Businesses guide hub.

Four search modes, one 700M+ profile index

Different recruiters think in different ways, so the platform supports four distinct ways to start a search. All four run against the same index and produce the same kind of ranked, scored output; the modes only change how you express what you want.

Plain English search

Type what you need the way you would say it to a colleague: "senior backend engineer in Bengaluru with Go and Kubernetes, 5+ years." The engine parses the intent, extracts the structured requirements, and runs the search. This is the fastest path from a hiring conversation to a candidate list, and it needs no training.

Paste a job description

If the role already exists as a written JD, paste the whole thing. CloudHire reads the description, identifies the title, seniority, must-have skills, and location signals, and converts them into a structured search. This mode is useful when hiring managers hand you a finished document and you want the search to reflect it exactly.

Boolean chip builder

Sourcers who grew up on boolean strings get full AND, OR, and NOT logic, but without typing syntax. CloudHire's chip builder assembles the query from clickable chips, so a complex nested expression is built visually and stays readable. There are no unbalanced parentheses and no silent syntax errors, which are the two classic failure modes of hand-typed boolean search.

Manual filters

For precise control, skip free text entirely and set structured filters directly: location, current and past titles, must-have versus nice-to-have skills, minimum years of experience, and education. Manual mode is also how you refine any search that started in another mode.

Conversational intake with live pool counts

A vague search produces a vague shortlist, so CloudHire's sourcing agent asks clarifying questions before it commits to results. If you search for "product manager," it will ask about seniority, the one skill that is truly non-negotiable, and the city or cities that matter. As you answer, CloudHire shows live pool counts, the number of matching candidates updating in real time with each answer. You can see immediately whether "must have fintech experience, must be in Pune" leaves you 12,000 candidates or 40, and loosen or tighten requirements before you ever open a results page. That feedback loop replaces the guess-search-revise cycle that eats hours on traditional databases.

Two editable Ideal Candidate Profiles anchor every search

From your intake answers, CloudHire generates two Ideal Candidate Profiles: structured pictures of who you are looking for, each with its own emphasis. Maybe one leans toward deep specialists and the other toward broader generalists with leadership signals. Both profiles are fully editable; change a skill, adjust the experience band, or rewrite the summary, and the search re-anchors around your edits. Having two explicit, visible profiles does something subtle but important: it makes the search criteria auditable. When a hiring manager asks why a candidate surfaced, the answer is written down in the profile rather than buried inside an opaque algorithm.

Cloud-ID blended ranking: verified candidates rise

Every result list on CloudHire blends verified and unverified candidates into a single ranking, with Cloud-ID verified candidates boosted upward. A Cloud-ID means the candidate's identity has been checked against a government ID and they carry a recorded, scored AI interview with them. Each verified candidate shows a badge directly on their row, and a "Cloud-ID only" filter narrows the list to verified profiles exclusively when identity assurance is the priority, for example in high-trust roles or when your team has been burned by fake applicants before.

The blended approach matters because verification should raise trust without hiding the rest of the market. You still see the full pool; you just see who is proven first. Read the dedicated guide to Cloud-ID verified hiring or the Cloud-ID product page for the full verification model.

Structured screening filters

Whatever mode you start in, results can be screened with structured filters:

• Location, down to specific cities
• Current and past job titles
• Skills, split into must-have and nice-to-have, so a missing must-have excludes a candidate while a missing nice-to-have only lowers their rank
• Minimum years of experience
• Education requirements

The must-have versus nice-to-have split is worth underlining. Most databases treat every keyword equally, which is how a candidate with nine of ten "requirements" but missing the one that actually matters ends up ranked first. CloudHire encodes that distinction explicitly.

An ATS match score on every candidate

Every candidate in a CloudHire result list carries a tiered ATS match score computed against your job, not a generic quality score. Click into it and the score expands to show the category weights behind the number, the specific skills that matched, the skills that are missing, and a written verdict summarizing the fit. Alongside the score, match insights flag red flags and warnings, the kind of signals an experienced recruiter checks manually: gaps, short stints, or title inflation relative to the role.

The practical effect is that screening starts before outreach. Instead of contacting fifty people and disqualifying thirty in the first call, you disqualify them in the results list, for free. The scoring model is the same engine described on the candidate scoring page, applied at sourcing time.

From results to outreach: views, collections, and contact reveal

Results are available in a card-style list view for reading depth and a table view for scanning volume, with bulk actions in both, so you can shortlist, save, or act on dozens of candidates at once. Saved collections let you organize candidates into named pools, a "Backend, Bengaluru, warm" collection today is a pipeline starter next quarter. Contact details, email and phone, stay locked until you choose to unlock a candidate, so you only spend reveal credits on people who cleared your screen. From there, candidates flow into engagement, covered in the guide to multi-channel candidate outreach.

Title recommendations and JD autofill

Two smaller features remove common friction at the start of a search. Title recommendations suggest adjacent and equivalent job titles you might not have thought to include, useful because the same role is titled differently across companies and markets. JD autofill drafts the job description fields for you from minimal input, so a search can be properly job-anchored even when nobody has written the formal JD yet.

Try it before you sign up

CloudHire lets you run a real search anonymously, before creating an account. You get genuine results from the live 700M+ index with candidate names redacted and contact details blurred. This is deliberately not a canned demo: you can judge pool depth, ranking quality, and match scoring for your actual role first, then sign up to unlock names, contacts, collections, and outreach. Independent feedback from teams already using the platform is collected on the reviews page.

On the roadmap

One sourcing capability is planned but not yet live, and it is worth being precise about that: semantic retrieval and re-ranking with per-candidate evidence chips. When it ships, ranking will lean further on meaning rather than keyword overlap, and each candidate will display small evidence chips citing exactly which parts of their profile drove their rank. Everything else described in this guide is live today.

Where sourcing fits in the CloudHire workflow

Sourcing is the front of a longer pipeline: candidates you find here can be invited straight into AI interviews with proctoring via magic links, scored, and tracked through offer. If you are comparing tools, the CloudHire vs LinkedIn Recruiter guide covers how job-anchored search differs from a connection-graph database.

Ready to see it on your own roles? Talk to the CloudHire sales team for a walkthrough with your real job descriptions.