This is the stack for the interview itself — the hour that produces the hiring decision and, at most companies, the worst record of it. It assumes you already source candidates and already run an ATS. The constraint it addresses is evidence: five interviewers, one candidate, and a debrief that runs on memory and tone of voice.
The question changed in July 2026. Greenhouse shipped Greenhouse Notetaker to Core, Plus and Pro — it joins the Zoom, Teams or Meet call, transcribes it, and maps AI-drafted notes to the scorecard questions the interviewer is accountable for, who then edits and submits. That is the baseline job the dedicated recorders in this category were built to do, and it now ships inside the ATS most readers of this page already pay for. On Greenhouse, the default answer is the bundled notetaker and no second contract. Everything below is what you buy when you can name the reason to spend past it.
The category also stopped being independent, which shortens the list further. Zoom announced its acquisition of BrightHire on 13 November 2025 and closed it that December; BrightHire sells under Zoom with its own brand, team, and domain intact. Pillar went first — Employ acquired it in March 2025 and rebranded it the AI Interview Companion, native inside Lever, Jobvite, and JazzHR. Metaview is the last independent of the three, and it has repriced away from this stack’s job. The practical consequence: the recorder you should buy is decided by your ATS and your conferencing platform, not by a feature bake-off.
If you want the whole hiring loop rather than the interview stage, that is the AI-augmented recruiting stack. If your problem is measuring technical skill rather than capturing judgment, that is the technical hiring stack. If your problem is proving the interviewee is the applicant, that is the hiring integrity stack. This one goes deep on one hour.
How the pieces fit
- Greenhouse is the rubric, the record, and now the recorder of first resort. Interview kits and scorecards are first-class objects, which is what makes the rest of this stack worth evaluating at all: a kit defines the competencies each interviewer is accountable for, and the scorecard is where the answer lands. Notetaker attaches the recording and the drafted notes to those competencies rather than to a video library nobody opens, and it sits in all three tiers — Core, Plus and Pro — so no upgrade buys it. Greenhouse is also the trigger surface: stage transitions tell the recorder which loop it is sitting in. Greenhouse MCP, announced in May 2026 and in open beta since 6 July across the same three tiers, makes the interview record readable by an approved agent under scopes a Site Admin sets in Dev Center → MCP Access, permission-aware and audited per call.
- BrightHire is the recorder when interviewer coaching or candidate fraud is the reason you are buying. Recording and transcription are the spine; Interview Planning drafts the plan before the loop opens, AI Interview Notes returns a filled scorecard, and BrightHire Screen runs an asynchronous first-round screen against a recruiter-defined rubric, sold standalone or bundled. Read the tier boundary before you price it: the Recruiters package is notes and 1-click scorecard completion, while the intake-and-debrief agent, the interviewer-coaching-tips agent, interview quality analytics, topic coverage analytics, deep ATS integrations, and SSO all sit in Teams. The REST API sits in Enterprises. Candidate Fraud Detection is generally available and spans applications, AI screens, and live interviews, flagging impersonation, coached answers, and multi-device cheating for human review rather than auto-rejecting.
- Metaview is the recorder when you are buying sourcing in the same purchase. Same core mechanics — joins the call, transcribes, writes a structured scorecard draft into Greenhouse, Lever, or Ashby — but the meter moved. Metaview now sells an agentic recruiting platform priced per user against profiles sourced: Free at $0 for the first 100 profiles, Pro at $100 per user per month for 200 profiles a month, Max at $300 per user per month for unlimited, Enterprise quoted. The notetaker rides along on every tier, so buying Metaview purely to record interviews means paying against a meter for work you are not doing.
- Pillar is the recorder if your ATS is an Employ product. Real-time question prompts, competency guidance, auto-generated scorecards, native inside Lever, Jobvite, and JazzHR. It belongs here as an alternative to the Greenhouse pairing, not a companion to it: on Greenhouse it is an integration competing with the ATS’s own bundled notetaker and two vendors that built for Greenhouse first.
- Zoom, Microsoft Teams, or Google Meet is the capture surface, the consent surface — and, for one job, a capability boundary. It is where the recording notice appears and where the recorder joins as a visible participant, which is the fact your candidates and your counsel will both ask about. It is no longer a neutral pipe on detection either: BrightHire’s deepfake detection runs on Zoom’s own models, so it is deepest on Zoom and explicitly reduced on Teams and Meet.
The handoffs that make it a stack
Each arrow is an event, not a habit.
A req opens and the recruiter builds the loop in Greenhouse — kits per stage, competencies per interviewer. That kit is the object the recorder reads: when Greenhouse advances a candidate to an interview stage, the scheduled event carries the kit’s competencies into the recorder, so the notes come back shaped like the scorecard instead of like a transcript. The calendar invite carries the recording notice and, where AI analysis is involved, the consent request — before the call, not in the first 30 seconds of it. The interview runs and the recorder joins the Zoom, Teams, or Meet call as a participant. Within minutes of the call ending a drafted scorecard lands on the Greenhouse candidate record, and the interviewer edits and submits it — the draft is not the submission, and the moment you allow it to be, you have automated the appearance of structure rather than the practice. The debrief then runs against submitted scorecards with timestamped clips behind the contested ones, which is the difference between “I didn’t get a strong signal” and a 90-second replay. Afterward two readers work the same corpus: the interviewer analytics that name who talks 70% of the time and who never asks their assigned competency, and — where you have enabled MCP — an agent answering a hiring manager’s question against the scorecards and notes under audited scopes, instead of a recruiter exporting a CSV.
Why this combination
Because interview intelligence is worth nothing without a rubric, and a rubric decays without evidence. Structured interviewing is the precondition here, not the output — where no competency is assigned per interviewer, recording produces a larger pile of the same unstructured judgment, at a per-seat price.
Two load-bearing rules. First: one recorder, and the ATS owns both the rubric and the trigger. Two recorders on one funnel produce two scorecard drafts and no canonical record, and interviewers will submit whichever one loaded faster. Letting the recorder own the rubric — defining competencies in the recording tool instead of in Greenhouse — puts your hiring criteria inside the layer you are most likely to replace, and this is a category where two of the three vendors changed owners inside 18 months.
Second, new since Greenhouse rebuilt its plan ladder: do not buy fraud detection twice. Greenhouse’s Real Talent fraud and spam detection sits in Plus and Pro and screens the application. BrightHire’s Candidate Fraud Detection screens the screen and the live interview. They catch different stages of the same problem, so name where your leak is — inbound application volume or the interview itself — and buy the one that covers it before adding the other.
What it costs
Budget $30K-$110K a year for a 50-250 hire loop, ATS included, with the ATS as the dominant line.
- Greenhouse publishes no prices across Core, Plus and Pro and meters per employee rather than per recruiter, so the bill scales with company headcount. Vendr puts the median contract at $26,587 a year across 873 purchases, against a range of $10,137 to $74,492, with buyers negotiating an average 16% off the first quote. SSO sits in Core, so security review does not push you up a tier. Voice AI, the two-way screening interview Greenhouse built on its May 2026 Ezra AI Labs acquisition, is sold outside the plan ladder — budget it as a line item, not a tier upgrade.
- BrightHire publishes no numbers either. Vendr publishes anonymized data from 57 purchases as of February 2026: a median of $18,000 a year, with $15K-$35K at 5-15 users and under 100 interviews a month, $35K-$75K at 15-50 users and 100-300 interviews, and $75K-$150K+ above that. Price the Teams package, not Recruiters, if coaching analytics are the reason you are here.
- Metaview publishes per-user monthly tiers — $0 free, $100 for Pro, $300 for Max, Enterprise quoted — metered on profiles sourced. Size it against your sourcing volume, not your interview volume, because that is the meter you will hit first.
- Conferencing is a cost you already carry. Treat it as sunk for the notes job. The one exception is deepfake detection, which is deepest on Zoom by construction, so if interview-stage identity fraud is the reason you are buying, the conferencing platform is part of the purchase rather than a background fact.
The unit that matters is the marginal cost per recorded interview, not the gross one. With Notetaker included in every Greenhouse tier, BrightHire’s $18,000 median does not buy recording — it buys the coaching analytics, the fraud detection, and cross-ATS portability on top of recording you already have. At 800 interviews a year that delta is about $22 an interview. Ask whether the named owner of interviewer coaching would pay $22 an interview for the analytics alone, because that is the actual question, and it is a harder one than the pre-2026 version.
Common variations
- Notetaker only, nothing else. On Greenhouse Core or above, run the bundled notetaker for two full quarters, then look at what you still cannot answer. If the gap is “which interviewers are miscalibrated,” you have a coaching purchase. If there is no gap, you have saved $18,000 a year and one vendor relationship.
- BrightHire on the coaching trigger. Move when you have a named owner for interviewer analytics and a standing monthly slot for them to review it. The premium over the bundled notetaker is largely that coaching layer, and it bills the same whether or not anyone reads it.
- Pillar instead, when you are on Lever, Jobvite, or JazzHR. Native beats integrated on those three, and the budget case is better. On Greenhouse it is the wrong pick, and that is the whole rule.
- Swap Greenhouse for Ashby. Ashby’s native analytics answer the interviewer-calibration questions without a reporting upgrade, and both BrightHire and Metaview support it. Choose it when you are selecting an ATS now rather than replacing a working one — and price the recorder as a real line item, since Ashby does not bundle the equivalent.
- Notes only, no AI screener. BrightHire Screen and Greenhouse Voice AI both run first-round interviews with no human present. That is a different purchase carrying different disclosure obligations, and buying the notes layer alone is the defensible starting point. Add the screener when your first-round volume, not your curiosity, forces it.
When this stack is the right pick — and when it isn’t
Pick it when you run 50-250 hires a year through multi-interviewer loops, your debriefs are the weak link, and you already have kits and scorecards defined. It is also the right pick when someone has asked you to show why a specific hire was made — a recording tied to a defined competency and a submitted scorecard is the only artifact that answers that question a year later.
Skip the paid layer in four cases. Below roughly 30 hires a year, the same recruiter usually sits in most of the interviews and the memory problem this stack solves is not yet real. If you are on Greenhouse and cannot name which analytic the bundled notetaker fails to produce, the second contract is buying a feeling. If your hiring is high-volume hourly, the interview is a screening question set rather than a judgment loop — that is the high-volume recruiting stack. And if your loop has no rubric, fix that first with the interview loop builder; recording an unstructured interview yields an unstructured recording with a transcript attached.
What this stack does NOT replace
- Sourcing and scheduling. Nothing here fills the calendar. Candidates have to arrive from somewhere, and the panel still has to be booked.
- The hiring decision. The recorder produces notes, clips, and a drafted score. Weighing them is the panel’s job, and the debrief stays human — see interview loop design and the debrief summary workflow for the shape that survives contact with a disagreement.
- Skills assessment. Capturing judgment well is not the same as measuring whether someone can do the work. That is a separate layer with separate vendors — the technical hiring stack.
- Identity assurance end to end. Fraud detection at the interview flags a suspicious call. It does not prove that the person who applied, the person who interviewed, and the person who receives the laptop are one human — that chain is the hiring integrity stack.
- Your consent and disclosure obligations. Recording consent is jurisdictional, and AI analysis raises the bar again. Illinois’ Artificial Intelligence Video Interview Act (820 ILCS 42) requires advance notice that AI will be used to analyze the interview, an explanation of how it works and the general characteristics it evaluates, and written consent — and it bars evaluating applicants who have not consented. On request you must delete the video within 30 days, including backups held by anyone you shared it with, which is a data-retention requirement your recorder’s default settings will not satisfy for you. Where AI ranks or scores candidates, NYC Local Law 144 and its bias-audit and notice duties attach to you, not to the vendor. Consult counsel on how either applies to your jurisdictions, run the AI interview compliance audit before the first recorded loop, and read AI screening bias before you turn on scoring.
- Quality-of-hire measurement. This stack tells you how the interview went and how the interviewer performed. Whether the people you hired turned out to be good at the job is measured after they start, against performance data no recorder holds.