Candidate screening

Candidate screening is an automation that reads each job application against the role's defined, job-relevant requirements, then organizes and summarizes the candidates and surfaces a shortlist with the specific evidence for each and any gaps flagged. It does the first-pass reading a recruiter would do by hand, at the same depth for every applicant, and hands back a structured view instead of a pile of resumes. The point is not to hire or reject anyone automatically. It is to put the reading in order so a person spends their time on the strongest candidates and can see exactly why each one is there.

Keyword filters match strings. A resume that says "client success" gets cut when the filter is looking for "customer success," and a keyword-stuffed application that has none of the actual experience sails through. This reads the application the way a person would, understands that two different phrasings mean the same skill, and evidences a candidate against what the role genuinely requires rather than whether they happened to use your exact words. It also shows its reasoning for every candidate, so you get a summary with evidence, not a pass-fail on a term match.

This is the part that has to be right. An AI screener that learns from past hiring or infers protected characteristics can discriminate at scale, which is a serious legal and ethical risk, and a biased or opaque screen is far worse than slow manual reading. So it is built to score only job-relevant criteria, never to infer or use protected-class signals, and to keep an audit trail of why each candidate was rated as they were. It never auto-rejects, and a human reviews everyone advanced or set aside. This is a regulated area, covered by things like EEOC guidance on discriminatory hiring, disparate-impact standards (outcomes that skew against a protected group even without intent), and rules such as NYC Local Law 144 on automated employment decision tools. This is not legal advice, and we recommend your HR and legal teams review the criteria and process before it goes live.

No. It never rejects or hires anyone. It reads, organizes, and shortlists, and it flags gaps, but every decision to advance or turn down a candidate is made by a recruiter or hiring manager who reviews the output, including the candidates it did not shortlist. It is decision support, and it is deliberately built so a person stays in the loop on everyone. If a tool is auto-rejecting applicants, that is a different and riskier thing, and it is not what this is.

On the applicant side, your ATS, such as Greenhouse, Lever, Ashby, or Workable, so it reads applications where they already land. On the output side, Slack, email, Google Docs, Google Sheets, and your CRM such as Attio, so the shortlist and summaries show up where your team works. Most tools with an API can be added. The audit maps your exact stack and connects to the systems you already run rather than adding another place to log into.

Usually two to three weeks. The first days go to defining the job-relevant criteria with you and your HR and legal review, then connecting your ATS. After that the screen runs against real applications and is bias-tested and tuned so the shortlist and reasoning hold up before anyone relies on it. The extra care compared with a simpler automation is deliberate, because this touches hiring decisions and has to be fair, auditable, and defensible from day one.

Two parts. Tooling runs as a modest monthly cost for the reading and model usage, which scales with how many applications you put through it. Implementation is a fixed scope, quoted once the audit maps your roles, criteria, ATS, and review process, so you are pricing a defined build rather than an open-ended retainer. The audit itself is where the scope and price get set against every other opportunity in your business, priced as part of that engagement, so you are not guessing at effort up front.