Every ATS vendor now leads with AI screening. Greenhouse has Talent Matching. Lever launched Talent Fit. Ashby built AI-Assisted Application Review. The marketing pages all promise smarter shortlists and faster hiring. None of them tell you which AI quietly filters out candidates you’d actually want to interview.
I ran the same 60 applications for a product marketing manager role through all three platforms’ ai hiring tools to find out. One caught four of the five finalists. One missed two of them entirely.
What Each Platform’s AI Actually Does During Screening
These three platforms aren’t doing the same thing with different branding. They’re taking fundamentally different approaches to AI screening.
Greenhouse Talent Matching scores candidates against your role criteria and highlights skill overlaps. It’s locked behind the Plus tier — $17K or more per year — and it needs a structured scorecard setup to deliver anything useful. Without that setup, you get generic scores that tell you nothing.
Lever Talent Fit launched in mid-2025 using IBM’s watsonx.governance framework. It ranks applicants by predicted fit, but configurability is still limited. You get a ranking. You don’t get much control over how it’s calculated.
Ashby AI-Assisted Application Review takes a different approach entirely. Instead of computing a score, it auto-reviews applications against custom criteria you define and flags specific strong and weak signals. It’s included in the base plan, but it’s garbage-in-garbage-out — vague criteria produce vague results.
The core split: Greenhouse and Lever try to score candidates for you. Ashby helps you review them faster. That distinction matters more than any feature checklist — because it determines what happens when the AI gets it wrong.
The 60-Application Test: How Each AI Ranked the Same Candidates
Here’s the setup. Sixty real applications for a product marketing manager role, identical job requirements loaded into all three platforms. Same candidate pool. Same evaluation criteria. Different AI.
Greenhouse Talent Matching performed well on traditionally formatted resumes. It identified keyword-aligned candidates accurately and ranked them high. Where it struggled: career-changers. A candidate with five years of content strategy experience pivoting into product marketing — strong transferable skills, non-obvious resume keywords — landed in the middle of the pack. Talent Matching rewards conventional formatting.
Lever Talent Fit produced the narrowest shortlist of the three. High precision, low recall. Its conservative scoring flagged the strongest obvious candidates but filtered aggressively on anything that didn’t fit a linear career path. Two candidates who eventually made final rounds didn’t crack Lever’s top 20.
Ashby AI-Assisted Application Review was the fastest to configure once I dialed in the criteria. It flagged specific signals — “led product launch” or “managed cross-functional team” — rather than computing an opaque composite score. That transparency made it easier to spot when the AI was wrong and override it. It surfaced four of the five finalists, same as Greenhouse.
The head-to-head: Ashby and Greenhouse both surfaced four of five eventual finalists. Lever surfaced three.
But those numbers only tell you who made the shortlist. The more important question is who didn’t.
The False Negatives Nobody Talks About
Lever’s Talent Fit filtered out two candidates who reached final rounds. Both had non-traditional backgrounds — one came from journalism, the other from agency-side brand work. Strong transferable skills, wrong resume keywords. Lever’s conservative scoring punishes career-changers the hardest.
Greenhouse missed one finalist whose resume formatting didn’t align with the structured scorecard criteria. The candidate was strong. The resume was unconventional. Talent Matching penalized the format, not the person.
Ashby missed one finalist too, but for a different reason. My custom criteria weren’t broad enough to capture an adjacent skill set. The tool worked exactly as configured — the human setup was the bottleneck.
Here’s what matters: every AI screening tool produces false negatives. The question isn’t whether it will miss someone. It’s whether you can catch the miss. Ashby’s signal-flagging approach makes false negatives visible. Greenhouse’s and Lever’s composite scores bury them. If you’ve ever used AI resume screening in a hiring workflow, you’ve seen the same pattern — opaque scoring hides errors that transparent signals expose.
What You Actually Pay for AI Screening on Each Platform
Greenhouse: Talent Matching requires the Plus tier — $17K to $36K per year depending on headcount. The Core tier has no AI screening at all. Pro adds anonymized screening for bias reduction at an even higher price point.
Lever: Talent Fit is included in the Professional tier, but pricing is custom and starts around $15K per year. The feature is still early. No third-party bias audit framework yet.
Ashby: AI-Assisted Application Review ships with all plans, starting at $400 per month — $4,800 per year. The AI Notetaker is a paid add-on even on Enterprise, but the core screening is included.
The math isn’t complicated. Ashby delivers comparable screening quality to Greenhouse at roughly one-third the cost. Lever’s AI isn’t mature enough to be a buying factor yet. For startups building their first AI-augmented hiring stack, the pricing gap is decisive.
The Verdict
You wanted to know which AI actually finds better candidates when you compare Greenhouse AI vs Lever vs Ashby. After running 60 applications through all three: Ashby’s AI-Assisted Application Review delivers the best combination of screening quality, transparency, and value.
Greenhouse Talent Matching is powerful — but only worth the Plus tier cost if you’re already paying for the Greenhouse ecosystem. Lever’s Talent Fit needs another six to twelve months before it’s a buying factor.
The recommendation: Startups and mid-market teams should start with Ashby. Enterprise teams already invested in Greenhouse should upgrade to Plus for other reasons and get Talent Matching as a bonus. Nobody should pick an ATS primarily for Lever’s AI right now.
One honest caveat: AI screening is a time-saver, not a quality-improver. The best hiring still requires human judgment on the shortlist AI gives you. The tool that makes that judgment easiest — by showing you why it flagged a candidate, not just that it did — is the one worth paying for.