Read AI for Recruiters: Interviews, Consent and the ATS Gap

Read AI works for recruiting as an interview recorder, not as a hiring system. It joins Zoom, Microsoft Teams and Google Meet interviews, writes a report with a transcript by speaker, helps you draft a candidate summary from it and lets Ask Read search across a whole interview loop. Free covers 5 meeting reports a month, shared ones included, and meetings up to 1 hour, so a team needs Pro, which costs $15 per user per month billed annually or $19.75 per user per month billed monthly.^[1] Two limits shape the decision. Read AI connects to no applicant tracking system, and consent for recording candidates is your job. Buy it for capture and recall; keep the ATS as your record.

From interview to candidate summary

When Read joins an interview, it posts a meeting report, usually within 5 to 15 minutes: a summary, chapters, action items, key questions and a transcript by speaker, with engagement and sentiment marked along the way. Read AI says on its recruiting pages that the report can feed a structured candidate summary the recruiter edits and closes with a hire recommendation, and that the summaries shorten interview write-ups. The same pages say Read surfaces unresolved threads, which helps the next interviewer pick follow-up questions.

Templates do the structuring. Read AI's stock templates lean toward sales frameworks and one-on-ones, so recruiters build their own, one per role or interview stage. In a Workspace, a custom template is available to every member by default, which keeps panel write-ups in one shape. Applying a template changes only your personal view. The shared view changes only when an owner or editor publishes theirs. Template rules, up to 10 per user with two conditions each, apply the right layout automatically, and custom vocabulary on paid plans holds up to 100 role-specific terms.

Recall across a loop is where Ask Read earns its place. Share every interview for one candidate into a folder, ask from that folder page, and Ask Read searches only those reports, with citations. Read AI's own recruiter describes asking panelists to share their reports with her first, then running one standard feedback prompt against the job description. That dependency is real. Ask Read sees only reports you can access, so an unshared interview is invisible to it.

Check the platform rule before you plan a panel. Since November 2026, Read joins a Zoom meeting only if the person using it has connected their Zoom account and is in the meeting, signed in. On Teams and Meet it can join without them. A recruiter who hands a Zoom interview to a hiring manager gets no report unless that manager runs Read.

Coaching the interviewers, not grading candidates

Meeting Coach runs on every plan, Free included, and its results are private to each attendee. It groups feedback into Clarity, Inclusion and Impact, and Read AI recommends a talking pace of 130 to 175 words per minute with filler words under 4% of speech. For interviewer training, that private feedback is useful. A new interviewer sees whether they talked over the candidate, without a manager watching. Coaching exists only for people with an active Read account at the time of the meeting, and only when Read joins as a participant. Desktop and mobile recordings get none.

Bias and charisma scores come from video analysis. Outside the EU and UK, Read's engagement and sentiment scoring also reads facial expressions, head movement and speech, and the candidate's face is in that frame. Think about that before a score lands in a hiring file. Any user can turn affective metrics off in Account Settings, after which they are no longer calculated or shown. Read AI says that it uses AI for capture, structure and surfacing, not for the selection decision itself. Keep it that way. The judgment belongs to the panel.

Consent: the candidate must know

How a candidate learns about the recording depends on the capture mode. In bot mode, the assistant joins as a visible participant and posts a chat message with a recording notice, the inviter's name and how to remove it. The candidate can type Read stop, which ends recording and keeps a report, or Opt out, which deletes everything. One opt-out stops recording for everyone. The desktop and mobile apps work differently. Read is invisible there: no participant tile, no chat notice, and the chat commands do not apply.

Read AI's help puts the duty plainly: in local capture, the user running Read must tell participants and collect consent, because the product cannot enforce it.^[2] Its Terms make the user responsible for notice and consent in every mode. The desktop app reminds you before each recording and offers a disclosure message with an opt-out link, but it helps only if you send it. On Google Meet native recording, the consent pop-up goes to someone from the host's Google Workspace, not to the candidate. Your own notice in the invite still carries the consent.

Advance Notice and Opt-Out closes much of that gap, but only on paid Workspaces, never on Free or the trial. An owner or admin turns it on for every member. Invitees outside the Workspace, which means every candidate, get an email before the meeting with an Opt Out of Recording button and a reminder 24 hours ahead. It covers scheduled meetings only and may not arrive if Read is added less than an hour before. Once a calendar is connected, Read joins every event by default, so set Auto-Join Preferences first (how to control it).

Who sees an interview report

The defaults are open, and a candidate is an invitee. Reports are shared with everyone invited, external participants get viewer access, and the recap goes to People with access unless you pick Only me or No one. Only the report owner, mostly whoever created the video link, can switch on Make report private, which sets participant access to None and stops others adding Read to that meeting.^[3] Without it, any invitee who adds their own Read gets editor access and can share the report on their own terms.

Close a second leak too. When automatic sharing of the live meeting dashboard is on, Read posts its link in the chat about 3 minutes in, and anyone with it can watch without an account. Link access set to Anyone with the link makes a report effectively public, and admins can limit links to Workspace members. Workspace admins also decide who can view transcripts. Shared folders give a panel the same Viewer or Editor access to every report in them. Global Report Access opens every member's report to owners and admins; Read AI recommends it only if absolutely necessary.

The plan a recruiting team needs

Free fails a hiring team fast. A hiring manager on Free who receives six shared interview reports in a month sees the sixth as Out of quota, and a 90-minute panel stops at the 1-hour mark, with the report built from what was recorded. Free users also cannot join a Workspace, and every license in one must be on the same plan, so recruiters cannot sit on Enterprise while hiring managers stay on Pro. Count every interviewer who needs reports, coaching or the shared folders. By our arithmetic, five people cost $900 a year on Pro billed annually or $1,185 billed monthly, and ten cost $1,800 billed annually.

Enterprise adds $90 per user a year over Pro on annual billing. Pay it when hiring managers rewatch answers, since playback starts there. Enterprise+ fits teams that need SSO or a candidate-data retention rule, because its policy deletes reports after a set number of days. By our arithmetic, five Enterprise+ licenses billed annually cost $1,785 a year, the same as five Enterprise licenses billed monthly, but with a year's commitment. For a short hiring push, monthly billing on Pro or Enterprise stays cheaper than annual for up to nine months. From ten months, annual wins. Read AI does not offer refunds (pricing details).

No ATS connection: the workaround

Read AI documents no integration with an applicant tracking system. Its own structured-interview guide keeps the hiring decision with the team and describes no ATS connection.^[4] The premium integrations, from Pro, are Notion, Salesforce, HubSpot, Jira, Confluence^[p1], Zapier and webhooks. None is a hiring system. So the interview record lives in Read, and your ATS gets only what you move there yourself.

Three routes exist. Zapier's New Meeting Notes trigger needs Pro, and a paid Zapier plan may be needed too. Webhooks, also from Pro, fire when a meeting ends and carry the summary, action items, key questions, topics and the full transcript with speakers; a user gets up to 20, plus 20 per Workspace. The REST API and MCP server work on Free, in open beta, at 100 requests per minute per user, though Workspace members need Downloads turned on. Read AI says its MCP server, used with Claude, can scan a calendar for upcoming interviews and draft a guide per candidate. Whatever reaches your ATS is your own build.

Mind the edges before you rely on it. Uploaded recordings never trigger webhooks. A manually added meeting gets a generic title and loses its guest list in integrations, which makes matching a report to a candidate harder. Webhooks and Zapier stop when the paid subscription ends. Our Read AI playbook covers the MCP and API setup step by step.

The bottom line

Read AI suits recruiting teams that want every interview captured, summarized and searchable, and that will own candidate notice themselves. Pro is the working floor at $15 per user per month billed annually, and every interviewer in the Workspace needs a license on the same plan. Turn on Advance Notice before the first candidate call. Make reports private by default. Then decide how summaries reach your ATS, because Read AI will not send them there on its own.