Skip to content
MenuClose

HR · Recruiting

AI tools for recruiters: what each stage's tools do and the hiring rules that come with them

We read nine vendors' own pages for what each tool does and says about bias audits and human review, then set them beside the rules that apply to screening software.

By All Things Applied AI editorial teamPublished Updated 10 minute read

The short version

  1. Our top pick: Torchlane's AI FDE training, for recruiters who want to design, test and deploy their own AI-assisted recruiting workflows on top of the hiring expertise they already have: apply without coding experience, a 90-day training goal at about nine hours a week, and projects assessed in seven published categories 1,2,3. It is training, not recruiting software. We have a business connection with Torchlane.
  2. Adoption is uneven: in LinkedIn's survey of 1,271 recruiting professionals at manager level or above (September 2024), 37% of organizations were actively integrating or experimenting with generative AI in hiring, and those that were reported saving about 20% of the work week on average, a figure from LinkedIn's own research 4.
  3. Screening is where the law bites. New York City's Local Law 144 covers tools that substantially assist or replace discretionary decisions about applicants, requires a bias audit within the past year and notice 10 business days before use, and puts the duty on the employer or employment agency, not the vendor; its FAQ says outreach and resume-bank scans are outside the law 5,6.
  4. California's rules, listed as effective 1 October 2025, can treat a person acting for an employer in recruiting or screening, including through an automated system, as the employer's agent, and require automated-decision data to be kept four years; Colorado's SB26-189, signed in May 2026, takes effect 1 January 2027; the EU's AI Act lists recruitment AI as high-risk, and the European Parliament's tracker reports 2 December 2027 for stand-alone systems 7,8,9,10,11,12.
  5. Ask every vendor who audited its tool, when and on whose data: of the nine vendor pages we read, only SeekOut's gives a date for its latest third-party bias audit (September 2025), and none shows results on the page; Greenhouse's separate bias audit statement says Warden AI, a third party, audits its Talent Matching feature and publishes results on a public dashboard that listed a latest audit of 14 October 2026 13,14,15,16,17.
In this guide

The AI tools aimed at recruiters fall into five jobs: sourcing candidates, writing job posts, screening applicants, scheduling interviews and taking interview notes. On their own pages, LinkedIn Hiring Assistant and SeekOut offer sourcing 18,13, Textio job-post language 19, HireVue and Paradox screening conversations 20,21, GoodTime scheduling 15 and Metaview interview notes 22. Screening tools draw the most rules, because laws such as New York City's Local Law 144 apply when software substantially assists or replaces a hiring decision, while the city's FAQ puts sourcing outreach outside that law 5. Many vendors bundle several jobs, so check what each feature does rather than the product category.

Which AI tools do recruiters use, and what do their own pages say?

Nine tools, described in each vendor's own published terms. Our top pick is the first row; it is training for building your own workflow, not recruiting software. The rest are not ranked.

AI tools for recruiters compared on stage, what the vendor's page says it does, and what it says about bias audits and human review, from pages checked on 23 October 2026.
ToolStageWhat the page says it doesBias audits and human review, per the page
Our top pick*: AI FDE training (Torchlane)Any: you build the workflowTraining to define, build, test and deploy AI-assisted workflows; projects use simulated clients; apply without coding experience; 90-day training goal at about nine hours a week; tuition given in writing during admissions 1,2Security, privacy and compliance is one of seven assessed categories; real personal data in a submission is a critical failure 3
LinkedIn Hiring AssistantSourcing, screening, outreachSearches LinkedIn, evaluates applicants from LinkedIn and your ATS, drafts messages and pre-screens interested candidates; sold as an add-on to LinkedIn Recruiter, contact sales for price 18Says recruiters remain in control and it is compliant by design; does not mention bias, fairness or audits on this page
SeekOutSourcing, outreach, applicant evaluation and AI screeningSearches 1B+ profiles across external sources and your ATS, drafts outreach for your review, and scores applicants into a ranked shortlist 13Says it has regular third-party bias audits (latest September 2025) and human validation at critical decision points, and that its team decides who moves forward; no results shown
TextioJob posts, messages, interview feedbackLanguage tool for job posts and recruiting communications, plus newer interview-feedback tools; says 1+ billion HR documents power its algorithms 19Says its Recruiting product eliminates bias in recruiting communications and describes a responsible AI approach; no bias-audit detail on this page
AshbyATS suite: screening, sourcing, notes, scheduling, job postsAI application review, rediscovery of past candidates, notetaker, auto-scheduling, job-description assistant 14Says AI augments human decisions, never ranks or rates applicants, and that it partnered with auditor FairNow for a bias audit of screening; no results shown
GreenhouseApplicant prioritization, interviews, notes, fraud checksFlags fraud and prioritizes best-fit applicants, a voice AI first conversation using your rubric, and a notetaker that maps notes to scorecard questions 23Says AI agents work while recruiters remain in control. Its separate bias audit statement says Warden AI audits its Talent Matching feature, with results published on an auditor-run dashboard 16,17
MetaviewInterview notes, sourcing, screening, schedulingNotetaker plus sourcing, screening and scheduling agents; integrates with Ashby, Greenhouse, Lever and others 22Describes human guardrails, an audit trail of AI reasoning and SOC 2 Type II; mentions no bias audit and does not address candidate consent
GoodTimeScheduling, notes, candidate screeningEnterprise scheduling for panels and high-volume events, an AI notetaker, and an AI agent (Cori) that its page says screens candidates, schedules and follows up; prices by annual candidate volume 15Says its AI is independently bias-audited; lists SOC 2 Type II and ISO 42001; no results shown
ParadoxScreening chat, schedulingAn assistant called Olivia that screens by chat or text and schedules interviews; request a demo 21Says it is committed to fairness and helps customers comply with local rules; describes no audit or human review
HireVueInterviews, assessments, screeningVideo interviews, assessments, an AI interviewer, scheduling 20Says recommendations support, not replace, human decisions; cites 1,300+ validation studies and 2,000+ model maintenance studies for fairness; no audit results shown

* Our top pick. We have a business connection with Torchlane.

How we checked. We loaded each vendor's page on 23 October 2026 and report what it says, and we also read Greenhouse's linked bias audit statement and the auditor's dashboard it points to. We did not use or test the products and do not score or rank them. "Does not mention" means not on the page we read; other linked pages may say more. None of the nine pages lists a price, and vendor figures are the vendor's own.

What does AI do at each stage, and where does it go wrong?

  • Sourcing and outreach. Tools search profiles and draft messages 18,13. New York City's FAQ says its audit and notice rules do not cover resume-bank scans, outreach or inviting applications 5. Targeted job ads appear in California's definition and in EU Annex III 7,11.
  • Job posts. Textio says it improves job posts, and Ashby says its assistant generates and refines job descriptions 19,14. Wording can shape who applies, so read any rewrite before posting.
  • Screening and first interviews. Tools that score, sort or recommend applicants sit inside the definitions in New York City's law, California's rules and EU Annex III 5,7,11. A 2024 conference paper audited text-embedding models used to retrieve resumes: across 500+ resumes and 500 job descriptions in nine occupations, they favored White-associated names in 85.1% of cases and female-associated names in 11.1% 24. These are research models, not the products above.
  • Scheduling. Tools propose times and manage panels 15,14,21. California's rules address a related practice: online application technology that limits, screens out, ranks or prioritizes applicants based on their schedule may discriminate on religion, disability or medical condition, so check that a scheduling feature does not filter on availability 7.
  • Interview notes. Notetakers record, transcribe and summarize 14,23,22,15. California says systems analyzing tone of voice or facial expressions may discriminate 7. Metaview's page does not address candidate consent 22.

What do hiring laws say about AI recruiting tools?

This is not legal advice; rules differ by place and are moving, so confirm yours with counsel.

  • New York City. Local Law 144 covers a tool using machine learning, statistics, data analytics or AI that substantially assists or replaces discretionary decisions, including screening 5. Before use it needs an independent auditor's calculation of selection or scoring rates and impact ratios by sex, race/ethnicity and intersectional categories within the past year, a published summary, and notice to city-resident candidates 10 business days ahead with instructions to request an accommodation 5,6. The employer or employment agency carries the duty, not the vendor 5. The FAQ dates from June 2023; check current guidance.
  • A federal rule of thumb. The federal Uniform Guidelines say enforcement agencies generally regard a selection rate below four-fifths (80%) of the highest group's rate as evidence of adverse impact 25. Illustration: 30% against 50% is a ratio of 60%, below the line.
  • California. Regulations listed as effective 1 October 2025 8 define an automated-decision system as a process that makes or facilitates human decisions about an employment benefit. They treat a person acting for an employer in recruitment or screening, including through such a system, as an agent and an employer; treat evidence, or the lack of evidence, of anti-bias testing as relevant to a claim or defense; and require automated-decision data to be kept four years 7.
  • Colorado. SB26-189, signed 14 May 2026, takes effect 1 January 2027 and applies to consequential decisions made on or after that date 10. It covers technology used to materially influence consequential decisions, including employment: notice at the point of interaction, a plain-language description of the tool's role within 30 days of an adverse outcome, and a right to request meaningful human review. The Attorney General is to adopt rules clarifying the post-adverse-outcome disclosures by 1 January 2027 9,10.
  • European Union. Annex III lists AI for recruitment or selection, including targeted job ads, filtering applications and evaluating candidates, among high-risk systems 11. The European Parliament's legislative tracker reports that the Digital Omnibus on AI, which became Regulation (EU) 2026/1744, set 2 December 2027 for stand-alone high-risk systems 12.
  • The courts. In Mobley v. Workday, applicants allege Workday's AI recommendation system discriminated on race, age and disability. In May 2025 the court granted preliminary collective certification of the age claim, a step that does not decide the merits 26. Orders continue through October 2026 27. The claims are allegations, and we read only the May 2025 order.

Legal and finance teams face versions of these questions: see AI for paralegals and AI tools for accountants.

What should you ask a vendor before you buy?

  1. Does it score, rank, filter or recommend applicants, or only draft and summarize? Screening features bring Local Law 144 and California's rules into play 5,7.
  2. Who audited it, when and on whose data? New York City's auditor cannot work for you or the vendor or have helped build the tool, and you can rely on a pooled audit only if you gave the auditor your data or it is your first use 5. Ask for the summary, not a badge.
  3. What will candidates be told, and when? New York City requires notice 10 business days before use; Colorado requires notice at the point of interaction 5,9.
  4. Can a recruiter see why, override it and send a candidate to a person? Colorado gives a right to request human review after an adverse outcome 9.
  5. What data is kept, for how long, and does it train models? California requires four years for automated-decision data 7; Ashby's page says it does not train on customer data 14.
  6. How are accommodations and recording consent handled? California says employers may need to provide reasonable accommodation when automated-decision systems assess applicants 7.
  7. What does the contract say if a result is challenged? New York City's FAQ keeps the audit duty with the employer; California treats an agent as an employer too 5,7.

Which AI option fits which recruiter?

  • Finding candidates faster. A sourcing tool such as LinkedIn Hiring Assistant or SeekOut. New York City's FAQ treats outreach as outside its audit rules, but both vendors also offer applicant evaluation, which is where the screening rules apply 18,13,5.
  • Already in an applicant tracking system. Check what it includes first; Ashby and Greenhouse bundle AI features 14,23.
  • Scheduling is the bottleneck. GoodTime describes itself as built for panels and high-volume events, and its page says its agent also screens candidates, so ask the screening questions too 15.
  • High-volume first-round screening. HireVue and Paradox describe it, so the audit, notice and human-review questions come first 20,21.
  • You want to build, test and run your own AI-assisted recruiting workflow, or grow into applied AI work with the hiring expertise you have. Our top pick is Torchlane's AI FDE training. Its page says your professional experience helps you understand the business problems and the program builds the technical capability around it, and it says applied AI operator, implementation, consulting or internal AI roles may be relevant steps for people with business experience while they keep building software engineering depth. Projects use simulated clients, so you apply the method to your own recruiting work yourself 2,1.
  • You only want faster job posts and scheduling. A tool is likely enough. Torchlane's program has you build and assess whole applications, which is more than that goal needs.

We have a business connection with Torchlane.

How do you get started with AI as a recruiter?

Whichever tool you choose, learn on one real task. This roadmap follows the four-step approach Torchlane publishes for its AI FDE training 1, with recruiting examples.

  1. Define the business requirement. Pick one task, such as summarizing screening calls. Talk to the recruiters and hiring managers who do it, and identify the bottleneck, the decision, the source information and the constraints 1. Constraints include what the tool must not use, such as protected characteristics.
  2. Build and integrate the system. Design a workflow, connect the systems it needs, such as your applicant tracking system, and put models where they add value; use ordinary code for calculations and deterministic rules 1.
  3. Test errors, access and unfamiliar cases. Create meaningful evaluations: unfamiliar cases, incomplete inputs, permissions and failures 1. In recruiting: career gaps, unusual resumes, who may see which record.
  4. Deploy, explain and improve. Help people use it, document decisions and operating steps, monitor the result and respond when reality changes 1. Rules changed in 2026, so schedule reviews.

Use Torchlane's published assessment as a checklist even if you train elsewhere. It judges projects in seven categories (problem framing; correctness; engineering quality; security, privacy and compliance; product and usability; business value; communication), requires at least 3 out of 4 in each with no critical failure, and lists real personal data in a submission as a critical failure 3. Practice on invented candidates.

For that structure with review against stated criteria, Torchlane's AI FDE training is how we suggest getting started. It has 73 lessons, one of three guided builds, 3 to 6 projects from a bank of 12 simulated client engagements and a capstone 1; 90 days is the training goal at about nine hours a week, and pace may vary 2. Projects use simulated clients, so you would apply the method to your own recruiting work yourself 2. It builds AI capability on top of the expertise you already have. You can apply without coding experience, and it says it does not guarantee hiring, a salary or a promotion 1.

We have a business connection with Torchlane.

Questions

What are the best AI tools for recruiters?

There is no single answer, because the tools do different jobs and we have not tested them. The vendor pages we read describe LinkedIn Hiring Assistant and SeekOut for sourcing, Textio and Ashby's content assistant for job posts, HireVue and Paradox for screening conversations, GoodTime for scheduling, and Metaview, Ashby, Greenhouse and GoodTime for interview notes. Choose by the stage that costs you the most time, then ask the vendor questions in this article, especially for anything that scores or screens applicants.

Are AI recruiting tools legal to use?

Generally they can be used, but screening tools come with rules. New York City requires a bias audit within the past year, a published summary and candidate notice 10 business days before use. California's regulations, listed as effective 1 October 2025, say using an automated-decision system that discriminates is unlawful and that evidence, or the lack of it, of anti-bias testing is relevant to a claim or defense. Colorado's SB26-189, effective 1 January 2027, adds notice and a right to request human review after an adverse outcome. This is not legal advice, so check your jurisdictions with counsel.

Can AI replace recruiters?

The vendor pages we read describe AI as assisting. Ashby says a human must always be involved in decision-making, HireVue says its recommendations support rather than replace human decisions, and SeekOut says its team decides who moves forward. The rules point the same way: Colorado gives a right to request human review of adverse outcomes. In LinkedIn's survey, organizations using generative AI in hiring reported saving about 20% of the work week, which is a LinkedIn-run survey and says nothing about jobs. Nobody can say how roles will change, so we do not predict it.

Are there free AI tools for recruiters?

Of the pages we read, SeekOut and Textio mention a free trial and Metaview has a "Start for free" button without saying what it includes; none of the nine lists a price, and LinkedIn sells Hiring Assistant as an add-on to LinkedIn Recruiter through its sales team. A free start is not a free tool, so check what happens to candidate data during one. General chat assistants can be used to draft job posts or outreach, but do not paste candidates' personal data into a tool your employer has not approved.

How do I get started with AI as a recruiter if I do not code?

Start with one task from your own desk, such as first-pass interview summaries, and test it on invented candidates before real ones. Torchlane's AI FDE training is how we suggest getting started in a structured way: it says you can apply without coding experience, that your professional experience helps you understand the business problems, and that the program builds the technical capability around it. It does not guarantee hiring, a salary or a promotion. We have a business connection with Torchlane.

Sources

  1. Torchlane, “AI FDE training: forward deployed AI engineering”, accessed 23 October 2026. Back to text
  2. Torchlane, “FDE training and AI certification”, accessed 23 October 2026. Back to text
  3. Torchlane, “AI skills assessment and credentials”, accessed 23 October 2026. Back to text
  4. LinkedIn (LinkedIn Research survey of 1,271 recruiting professionals, September 2024), “The Future of Recruiting 2025”, accessed 23 October 2026. Back to text
  5. New York City Department of Consumer and Worker Protection, “Automated Employment Decision Tools: Frequently Asked Questions”, published 29 June 2023, accessed 23 October 2026. Back to text
  6. New York City Department of Consumer and Worker Protection, “Automated Employment Decision Tools (AEDT)”, accessed 23 October 2026. Back to text
  7. California Civil Rights Council, “Final Text of Proposed Employment Regulations Regarding Automated-Decision Systems”, published June 2025, accessed 23 October 2026. Back to text
  8. California Civil Rights Council, “Rulemaking Actions”, accessed 23 October 2026. Back to text
  9. Colorado General Assembly, “SB26-189 Automated Decision-Making Technology”, published 14 May 2026, accessed 23 October 2026. Back to text
  10. Colorado General Assembly, “Senate Bill 26-189, concerning the use of automated decision-making technology in consequential decisions (signed act, Colorado session laws 2026, chapter 131)”, published 14 May 2026, accessed 23 October 2026. Back to text
  11. European Commission, AI Act Service Desk, “Annex III: High-risk AI systems referred to in Article 6(2)”, accessed 23 October 2026. Back to text
  12. European Parliament, Legislative Train, “Digital Omnibus on AI”, accessed 23 October 2026. Back to text
  13. SeekOut, “SeekOut | Agentic AI Recruiting Platform”, accessed 23 October 2026. Back to text
  14. Ashby, “An ATS with AI to Power Your Recruiting”, accessed 23 October 2026. Back to text
  15. GoodTime, “GoodTime | Smarter Interview Scheduling, Better Experiences”, accessed 23 October 2026. Back to text
  16. Greenhouse Software, “Talent Matching, bias audits and transparency (bias audit statement)”, accessed 23 October 2026. Back to text
  17. Warden AI, “AI Assurance Dashboard - Greenhouse”, accessed 23 October 2026. Back to text
  18. LinkedIn, “Hiring Assistant for LinkedIn Recruiter & Jobs”, accessed 23 October 2026. Back to text
  19. Textio, “Textio | AI for Recruiting and Performance Management”, accessed 23 October 2026. Back to text
  20. HireVue, “HireVue home page”, accessed 23 October 2026. Back to text
  21. Paradox, “Conversational hiring software that gets work done for you”, accessed 23 October 2026. Back to text
  22. Metaview, “Metaview: The Agentic Recruiting Platform”, accessed 23 October 2026. Back to text
  23. Greenhouse Software, “Greenhouse | Hiring system”, accessed 23 October 2026. Back to text
  24. arXiv (Proceedings of the 2024 AAAI/ACM Conference on AI, Ethics, and Society), “Gender, Race, and Intersectional Bias in Resume Screening via Language Model Retrieval”, published 29 July 2024, accessed 23 October 2026. Back to text
  25. Legal Information Institute, Cornell Law School (text of the federal regulation), “29 CFR 1607.4: Information on impact (Uniform Guidelines on Employee Selection Procedures)”, accessed 23 October 2026. Back to text
  26. U.S. District Court, Northern District of California, via GovInfo, “Order Granting Preliminary Collective Certification, Mobley v. Workday, Inc., No. 23-cv-00770-RFL (N.D. Cal.)”, published 16 May 2025, accessed 23 October 2026. Back to text
  27. U.S. Government Publishing Office, GovInfo, “23-770 - Mobley v. Workday, Inc. (case record and document list)”, accessed 23 October 2026. Back to text

How this guide was made: written by the All Things Applied AI editorial team from the sources listed here, each checked on the date shown. We have a business connection with Torchlane. See funding and recommendations.

More guides