> This is the markdown version of https://www.classet.ai/blog/classet-vs-indeed-ai-screener > Learn more at https://www.classet.ai ![Comparing Classet's AI phone screen with Indeed's Smart Screening inside the application](/_next/image?url=%2Fimages%2Fblog%2Fclasset-vs-indeed-ai-screener.png&w=3840&q=75) # Classet vs Indeed AI Screener: The Candidate Problem Indeed's Smart Screening runs an AI interview inside the application. Here's what it screens, what it doesn't, and why the candidate backlash is about format and disclosure rather than AI. [![Paul Jones](/_next/image?url=https%3A%2F%2Fassets.basehub.com%2Fe0b5701f%2F6599306507912123f90f150a8bfaaf6c%2Fscreenshot-2026-01-28-at-10.53.16-am.png%3Fwidth%3D100%26height%3D100%26quality%3D100&w=96&q=75) Paul JonesHead of Growth at Classet ](/blog/authors/paul-jones) September 16, 2026 AI Recruiting **Quick answer:** Indeed Smart Screening runs an AI conversation inside the Indeed application and returns a 0–100 Smart Fit Score to your ATS. It's good at what it does, and it only does it for applicants who come through Indeed Apply. Classet calls every applicant, from any source, within seconds and runs a structured phone screen you control end to end. The candidate backlash that both tools get blamed for is not about AI: the top four reasons US job seekers abandon a process are format, disclosure, monitoring, and having no opt-out. Indeed now runs an AI interview inside the application itself, and the reasoning I hear from hiring teams is hard to argue with: the spend is already there, so why not see how far it gets us. That's a fair read. Indeed's Smart Screening is real, it's competent, and turning it on costs you an afternoon. Test it before you buy anything else, including ours. That reasoning hides a swap, though, because "AI screening" is not one product category. Indeed built a filter on Indeed's traffic, and we built a screening layer that belongs to you. Those two things look identical on a feature grid and behave nothing alike six months in, especially now that candidates have started keeping private blacklists of employers who use these tools badly. ## What is Indeed Smart Screening? Indeed Smart Screening is an AI screening product that interviews applicants inside the Indeed application and returns a 0–100 Smart Fit Score to the employer's ATS. It sits inside Indeed Apply. A candidate submits an application, and instead of dropping into your ATS as a resume, they go straight to what Indeed calls AI Recruiter: an adaptive question-and-answer conversation in text, chat, or video, depending on which formats you enabled. The output lands in your ATS as a Smart Fit Score from 0 to 100, scored against criteria you write in plain language, plus a summary of the candidate's strengths and gaps and a full transcript of the conversation. Indeed's own numbers on it: 54% fewer applications needed per hire and a 20% cut in time to hire. Madeline Laurano at Aptitude Research read the launch as a structural move rather than a feature release, and I think she's right. Indeed spent fifteen years being the place volume comes from, and screening at the point of application is how it stops being blamed for the volume. Indeed also ships the opposite bet at the same time. Interview on Demand puts a live human recruiter on video with an applicant, often inside four minutes. That product exists because Indeed knows exactly what the AI conversation costs them in candidate goodwill. ## The coverage problem nobody demos Smart Screening screens applicants who arrive through Indeed Apply. Read that again if you run multi-source hiring, because it's the entire argument. Nothing else touches Smart Screening: not your referrals, not your career-site applicants, not your Facebook group posts or your ZipRecruiter spend or the guy your foreman texted. You get a fast, structured, scored pipeline for one slice of your funnel and your old manual process for the rest, which means two standards of evidence for the same job and no clean way to compare a referral against an Indeed applicant. Then there's the part operators feel later. The screening record is Indeed's. The scoring model is Indeed's. The criteria are yours, but the weighting that turns those criteria into a 76 instead of a 61 is not something you can inspect, tune, or take with you when you move spend to another board. You are renting your top of funnel from the company that also sells you the traffic into it, and both prices are set by the same vendor. Ownership sounds abstract until the quarter you cut Indeed spend by 40% and discover your screening capacity dropped by 40% too. ## Candidates aren't refusing AI. They're refusing four specific things. Here's where the "candidates hate it" worry usually gets aimed at the wrong target. Greenhouse surveyed 2,950 job seekers this spring. 63% had been interviewed by AI, up 13 points in six months. 38% of US candidates have already withdrawn from a hiring process because it included an AI interview. CNBC ran two pieces this week on job seekers who now keep private blacklists of employers that use them. So the reflex is: AI interviewing is the problem, and my candidates will hate me for it. Look at what those candidates named. Reason candidates walked away Share of US candidates Pre-recorded video scored by AI, with no human present 33% Employer never disclosed how AI would be used 27% AI monitoring during the process 26% A required, AI-led interview with no alternative 26% Source: Greenhouse 2026 Candidate AI Interview Report, US respondents. 38% of US candidates have already withdrawn from a hiring process because it included an AI interview, and another 12% say they would. The four reasons US candidates gave were AI-scored recorded video with nobody watching, an employer who never mentioned AI, monitoring during the process, and a required AI interview with no way around it. Every one of those is a decision the employer made, and none of them is "a machine talked to me." The rest of the survey says the same thing from the other direction. Only 19% of US candidates want _less_ AI in hiring. 46% want the option to request a human instead, which most won't take because what they want is to know it exists. 38% want confirmation that a person reviews the AI's output before a decision. And 70% say nobody told them AI would be evaluating them until they were already in it. Then the number that should embarrass all of us: of candidates who completed an AI interview, 51% never got an outcome. Not a rejection. Nothing. A scoring model does not fix that. You can score a thousand applicants in an hour and still ghost every one of them, and the candidate's memory of your company will be the ghosting, not the score. ## Format is most of the argument If the top walk-away trigger is AI-scored video with no human present, then the format you pick is doing more work than the model behind it. Indeed defaults toward the application window: text, chat, or video, completed right there on whatever device the person applied from. That's smart for completion rates, because Indeed's own data says reaching a candidate inside 15 minutes holds connection odds near 85%, versus about 35% when you wait. We took the same speed and put it on a phone call instead. Joy, our AI recruiter, dials the applicant within seconds of the application landing. No app, no camera, no typing on a cracked phone screen between deliveries, no upload. A CNA finishing a double shift answers her phone in the parking lot and talks for six minutes. That same person is not filling out a chat interview at 11:40 PM, and she is definitely not recording video of her face. The format difference shows up as a disability and access difference too. One of the CNBC interviewees has nystagmus, a condition that makes her eyes move involuntarily; the video system kept interrupting to tell her to look straight ahead. She couldn't answer the questions. A phone call has no gaze to police. We wrote more on this tradeoff in [AI phone screens vs video interviews](/blog/ai-phone-screen-vs-video-interviews). ## What our own candidates said Classet collected 10,167 post-interview ratings from candidates Joy screened between January and August 2026, and 88% rated the experience positively. I'd rather argue candidate experience from that than from somebody else's survey. We read all 1,072 written comments attached to the negative ratings, one at a time, and categorized every root cause. The single biggest cause was call infrastructure: echoey speakerphones, dropped connections, dead microphones. Strip out only the ratings where the candidate described a dropped call or their own device failing, and the positive rate goes from 87.5% to 90%. Explicit anti-AI sentiment accounted for 5.8% of negative comments, which works out to 0.6% of all candidates. Thirteen people, 0.13%, asked for a human instead. The full breakdown is in [We read every negative AI interview review](/blog/ai-interview-negative-reviews), and the headline analysis is in [part one](/blog/ai-interview-candidate-experience). I'd rather you read the negative half. The market withdraws at 38% and 0.13% of our candidates ask for a human. That gap comes from the format, from disclosing the AI up front, and from a Classet screen ending in an outcome rather than silence. ## When Indeed's screener is the right call Turn it on, and don't book a demo with us, if: - Effectively all of your applicant flow comes through Indeed Apply already. - You want the applicant pile ranked rather than screened, and your recruiters will still call everyone near the top themselves. - You have one or two openings and nobody whose week is consumed by screening. - Your roles are salaried and desk-based, where a typed conversation is a fair medium for the candidate. Come talk to us when your applicants arrive from five places and you need one standard applied to all of them, when your roles are frontline or licensed and the screen has to verify a CDL or an active state license before anyone's time gets spent, when people apply at 11 PM and your competitor calls first, or when you want the transcripts, the recordings, and the scoring logic to still be yours after you renegotiate your job board contract. Indeed Smart Screening Classet **Who gets screened** Applicants who arrive through Indeed Apply Every applicant, from any source or job board **Default format** Conversational text, chat or video after applying Outbound phone call, answered on any handset **Time to first contact** Immediately after the application is submitted Within seconds of the application, 24/7 **Output** Smart Fit Score (0-100), summary and transcript in your ATS Transcript, recording, structured summary and knockout flags **Who owns the criteria** You define the criteria; Indeed owns the scoring model You define the questions, knockouts and scoring logic **Languages** Not published per-format 25+ languages on the same phone screen **Portability** Tied to Indeed traffic and an Indeed Connect subscription Runs standalone or on your ATS; the record is yours **Published pricing** Bundled with Indeed Connect / sponsored spend From $249/month standalone; from $1,995/month ATS-integrated Indeed details come from Indeed's own Smart Screening documentation and product announcements. "Not published" means the vendor does not state the figure publicly. ## Common questions Can I use Indeed Smart Screening and Classet together? Yes, and some teams do during an evaluation period. The thing to watch is double-screening: if an Indeed applicant completes an AI Recruiter conversation and then gets a call from Joy twenty seconds later, you've made the experience worse than either tool alone. Pick one system of record for screening and route the other source into it. Does Indeed's Smart Fit Score replace a phone screen? Indeed says explicitly that it doesn't. The score is built to prioritize review rather than make decisions, and Indeed tells employers to review flagged candidates and disregard the summary where it's wrong. Treated as a ranking signal it's useful. Treated as a pass/fail gate it will reject people whose resume parsed badly rather than people who lack the qualification. Will candidates drop off if I add an AI screen? Some will, and how many depends almost entirely on choices you control. The Greenhouse data says the drop-off triggers are AI-scored recorded video with no human, undisclosed AI, monitoring, and no alternative path. Disclose it in the job post, say what's being measured, name the human who reviews the output, and offer a human option nobody will use. On our own volume, 88% of candidates rated the screen positively across 10,167 ratings. What does Classet cost compared to Indeed's screener? Indeed bundles Smart Screening with Indeed Connect and sponsored spend, so its cost is inside your job board budget rather than a separate line. Classet publishes its pricing: from $249/month standalone, and ATS-integrated deployments start at $1,995/month across Greenhouse, Workday, Lever, Bullhorn, iCIMS, Workable, SmartRecruiters, and Ashby. Teams without an ATS can run Joy on our own [lightweight applicant tracking system](/product/joy-ai-powered-ats). Does Classet only work for hourly and frontline roles? It's built for them. A structured voice screen verifies availability, licensure, drive time, motivation, and baseline qualification, which is most of what decides a frontline hire and almost none of what decides a senior engineering one. For a wider view of the category, see our [candidate screening software comparison](/blog/candidate-screening-software). ## Where this leaves you Starting with Indeed's screener is the right call for most teams this quarter. It will take work off your recruiters the day you enable it, and you should find that out firsthand. The question that shows up in month four is narrower: when you look at your funnel, how much of it did Smart Screening ever see? If the answer is 40% of your applicants, you have a screening process for 40% of your applicants, and the other 60% still lands on a recruiter's desk as a resume. If you want to hear what a phone screen sounds like before you decide, [book a demo](/demo) and we'll run a live one on a role you're hiring for right now. Bring the Indeed transcripts and compare them side by side. That will tell you more than anything I can write here. [![Paul Jones](/_next/image?url=https%3A%2F%2Fassets.basehub.com%2Fe0b5701f%2F6599306507912123f90f150a8bfaaf6c%2Fscreenshot-2026-01-28-at-10.53.16-am.png%3Fwidth%3D100%26height%3D100%26quality%3D100&w=128&q=75) Paul Jones Head of Growth at Classet Paul comes from an operator background running an Alpine-owned company, and brings firsthand experience with the hiring challenges Classet was built to solve. He's driven by a belief that the right technology can make meaningful work more accessible. ](/blog/authors/paul-jones) Follow Classet in Google Add classet.ai as a preferred source and our hiring research is more likely to show up in your Top Stories and AI Overviews. [Add as preferred source →](https://www.google.com/preferences/source?q=classet.ai) ## Explore More ### Use Cases - [RPO / BPO Recruiting](/use-cases/call-centers-bpo) - [Data Centers Recruiting](/use-cases/data-centers) - [Healthcare Recruiting](/use-cases/healthcare) ### Integrations - [Greenhouse](/integrations/greenhouse) - [Bullhorn](/integrations/bullhorn) - [Lever](/integrations/lever)