Read AI for interviews: capture options, engagement scores and what to switch off

Read AI can transcribe and summarize job interviews on Zoom, Teams and Google Meet, either as a participant in the call, without a bot on Google Meet, or through its desktop app. As of September 2026 it is priced from free to $39.75 a month per user and names no applicant tracking system integration. The part that needs thought for hiring is Read's meeting analytics: engagement and sentiment measures designed for meetings in general, which should not be used to judge candidates.

This page covers the capture options, the plans, a settings checklist for candidate calls, and who Read suits. For similar general meeting tools, see tl;dv for recruiting and Fireflies for recruiting.

How Read AI captures an interview

Sources, as of September 2026: Read's article on stopping Read from joining meetings, its botless Google Meet announcement dated March 26, 2026, and the desktop app page. Read also offers a mobile app per its pricing page.

What the chat commands mean for candidates

When Read joins as a participant, Read says any participant, not only the organizer, can type "Read Stop" to make it leave with a report of what was captured, or "Opt Out" to remove it and delete the recording with no report. That is a useful protection for candidates. It also means an interviewer should expect that a candidate may use it, and have a plan for taking notes by hand if they do. Nothing about the candidate's choice belongs in the evaluation.

Which method to use

• Interviews you host on Google Meet: botless capture keeps the screen clear while still showing a notice before recording.
• Interviews on Zoom or Teams you host: the participant method gives candidates a visible notice and their own stop control. The desktop app avoids a participant, but notice becomes entirely your job.
• Interviews someone else hosts: the desktop app, with the host's agreement. A participant needs admitting and may surprise the host.
• In-person interviews: the desktop or mobile app, with a spoken notice at the start.

Read AI plans, in hiring terms

From Read's plans and pricing page, as of September 2026. Annual prices reflect the 25% discount Read shows. The last column is arithmetic.

Three details change the choice for interviewers:

• Free caps meetings at one hour. A 60-minute interview that overruns loses the end, which is usually where logistics and candidate questions are.
• Playback is Enterprise. If a colleague needs to hear an answer, not just read it, Pro is not enough.
• Custom retention is Enterprise+. Teams with a written retention period for interview records should check what the lower plans allow before rolling out.

Engagement and sentiment: handle with care in hiring

Read's recruiting page describes post-interview reports and measuring "moments and topics that increased engagement, or decreased sentiment". Those measures make sense for sales calls and team meetings. Applied to candidates, they carry real risk:

• They are not job requirements. How engaged a candidate appeared on video tells you little about whether they can do the work.
• They can track protected characteristics. Eye contact, expressiveness and speaking pace vary with disability, neurodivergence, culture, language and nerves.
• They may bring you into AI hiring rules. If a score substantially assists a hiring decision, rules such as NYC Local Law 144 or the Illinois AI Video Interview Act for recorded video analysis may apply. Take advice before using them.
• They are discoverable. A report saying a rejected candidate scored low on sentiment is an awkward document to explain.

A defensible middle ground: use analytics only for interviewer coaching (talk time, interruptions, whether the interviewer asked every question), never for candidate evaluation, and write that rule down. Our guide on reducing bias in interviews explains the wider approach.

Settings checklist for candidate calls

Invite text to copy

This interview will be transcribed with Read AI so I can focus on our
conversation. The notes are used only for your application with
[Company]. If you would rather not be transcribed, reply to this
invite or tell me at the start and I will take notes by hand. Your
choice will not affect your application.

Say it again at the start of the call. For states that require all-party consent, use the stricter wording in our interview recording consent script.

Turning a Read transcript into a scorecard

Read gives you a summary and a transcript. The scorecard is still yours to write, and the order you do it in decides whether the transcript reduces bias or just decorates a gut feeling.

1. Write your ratings from memory first, within the hour, one line of evidence per criterion. Do not open the summary yet.
2. Open the transcript, not the summary, and find the answer behind each rating. A summary compresses; a transcript shows whether the candidate said "we" or "I", and whether the example had an outcome.
3. Replace your evidence line with a short quote where the transcript supports it. Where it does not, lower your confidence or the rating.
4. Check logistics against the transcript: pay expectations, notice period, location. These are the details summaries most often round off.
5. Ignore the engagement view. If you have already looked, note that you did and rate only from the answers.
6. Submit, then share. Send the report to others only after your scorecard is in, so they cannot anchor you and you cannot anchor them.

The evidence-based interview feedback guide shows what a good evidence line looks like, and the interview scorecard template gives you the form.

Where Read AI stops for recruiting

Not found on Read's pages as of September 2026. Ask Read, since features change.

• No named ATS integration. Greenhouse, Lever, Ashby, Bullhorn and JobAdder are not on the pricing page's integration list.
• No interview kit or scorecard tied to your criteria described on the recruiting page; Read's articles on structured interviews are guidance, not a documented feature set.
• No submittal output for agency recruiters.
• Phone screens are not described on the desktop page, which names meeting platforms and in-person conversations. Test a softphone call first.

Who Read AI suits

A sensible choice if:

• Your company already uses Read for meetings, and interviewers want transcripts at no extra cost.
• You interview mainly on Google Meet and want a clear screen with a notice before recording.
• You want candidates to have their own stop control during the call.
• You will use its analytics for interviewer coaching only.

Look elsewhere if:

• You need notes in your ATS without Zapier.
• You are an agency recruiter whose output is a client submittal.
• You want a question guide during the interview and scores tied to evidence.

Interview Signal is built around that last point: a desktop app with no bot that tracks must-ask questions while the interview runs and writes a scorecard where each score carries quotes checked against the transcript. It has no engagement or sentiment scoring and no video, so teams that want playback and highlights may prefer Read's Enterprise plan.