Built This Week/AI by Industry

AI by Industry

AI in Recruiting and HR

AI in recruiting works best as extra capacity: score every resume, interview at any hour, coach every recruiter. Our recruiting team grew after we added it, and the human still makes the call.

Recruiting is where AI changed our own company the most. Ryz Labs' recruiting team runs thousands of interviews a year, so most of our hiring builds came from our own pain: an AI resume scorer, an AI avatar interviewer, a recruiter coaching tool and a video highlight reel builder. The pattern held every time: AI handles volume at the top of the funnel and the recruiter keeps the decision. On the HR side, Jennifer Dulski of Rising Team showed how AI can coach managers, with strict limits on what it will touch.

You know, in in the last thirty days, we've we've increased our recruiting capacity, our recruiters by 25%.

— Sam Nadler, Episode 6

How companies we talked to use AI in recruiting and HR

Rising Team: an AI coach for every manager

Jen spent about 25 years at Yahoo, Google, Facebook and Change.org, where she got executive coaches while frontline managers got nothing that scaled. Rising Team started in 2020 as team-building software; once LLMs arrived, it added RT, an AI "personal trainer" for leadership. RT runs an intake, builds a one-of-one growth plan, schedules coaching and role-play sessions on your calendar, and drafts scripts for hard conversations based on how each teammate likes feedback. She reports an average session rating of 8.5 out of 10 and a 10 to 33% lift in retention.

For HR teams, insights roll up anonymized, sliced by level or location but never down to a team. RT will never help with a firing, hiring or rating decision, and stricter companies such as financial firms add their own guardrails.

And the BCG data says 70% of whether your AI transformation is successful or not lands on whether the people are able to adapt to these tools and still work with each other.

— Jennifer Dulski, Episode 52

What we built

  • Ryz Score: popular roles drew more than 100 applicants each (152 for one Ruby on Rails job), and recruiters spend under 60 seconds per resume. Supabase edge functions send each resume, application and job description to ChatGPT with a scoring matrix for skills, experience, education and culture fit, and the score and rationale land back in the ATS. Recruiters rarely open the app itself.
  • NTRVSTA: after resume screening, interviewing was the big time sink, at 30 to 60 minutes each. Carolina, the AI avatar, runs dynamic, multilingual interviews any time and asks follow-ups. Candidates go from resume score to interview within minutes instead of days, get an email with their score, and adapt to the avatar quickly.
  • Sam's ATS: Sam built an applicant tracking system front end in about an hour with Claude Code, running three agents in parallel from two PRDs, to match exactly how our recruiters work.
  • ScreenEval: our recruiters run dozens of 20 to 25-minute screens a day. Sam's tool takes the call transcript, asks the recruiter for their own assessment first, then has Opus 4.5 evaluate the candidate and coach the recruiter on a six-area rubric with items like rapport, follow-up probing, compensation, next steps and pacing. Transcripts become searchable, for example for every candidate who mentioned Python. It took about six hours, from a host who had never written code.
  • Candidate highlight reel builder: AI reads an NTRVSTA interview transcript and proposes nine clips across English communication, role fit, and personality and drive. The recruiter trims about three and renders a branded reel with Remotion. Rendering takes five to ten minutes, versus hours or days with a video editor.
  • AI compliance training platform: annual HR trainings like harassment prevention, with an AI-organized curriculum and scenario videos made in Veo 3.1 Fast. About 60 to 70% of the video content was right on the first pass. Jordan framed it as build versus buy: cheap to build, expensive to rent per seat.

What's still hard

  • Trusting the score. We calibrated Ryz Score against recruiters' judgment before relying on it, and ScreenEval is built so the AI reinforces the recruiter's call instead of overriding it.
  • Fear of replacement. Sam said the team was apprehensive about NTRVSTA at first. Human interviews still happen later in our process, and the recruiting team grew.
  • Keeping an AI interviewer on track. Jordan said it took a lot of prompt engineering and red-teaming to keep Carolina friendly, on topic and appropriate.
  • Volume cuts both ways. Jen noted AI makes applying so easy that a job can draw 10,000 applicants, and sorting and interviewing them is a huge time and money sink. That makes retention worth more.
  • Rendering and polish. Getting the highlight reel to render in a reasonable time was the hardest part of that build.

Episodes on AI in recruiting and HR

FAQ

How is AI used in recruiting?

Our recruiters use it at every step: Ryz Score rates each resume against the job description inside the ATS, NTRVSTA's AI avatar runs first-round interviews, ScreenEval coaches recruiters on their phone screens, and a highlight reel builder cuts interview video into a short candidate clip.

Will AI replace recruiters?

Not in our experience. After adopting AI interviews, our recruiting capacity grew 25% in 30 days and we hired more recruiters, because each one could take on more roles. Human interviews still happen later in the process.

Can AI score resumes accurately?

Ryz Score sends the resume, application, job description and a scoring matrix to ChatGPT. Before trusting it, we tested whether a 90 or 92 actually felt like a 90 or 92 to our recruiters.

How is AI used in HR beyond hiring?

Rising Team's AI coach, RT, learns a manager's team through team-building sessions and helps rehearse hard conversations. By design it will never help with a firing, hiring or rating decision. We also built our own compliance training platform with AI-generated scenario videos.