​

Purpose

What has changed in Exterview, newest first, grouped by month. Looking for something else? Now Available lists everything you can switch on today. Upcoming and Roadmap cover what is still to come.

​

September 2026

​

v1.44.0: 11 to 24 September 2026

AI Employees arrive, Smaya answers stream in, the Create Job wizard gains a publish step and a view of the job’s pipeline, and Reports is now the Decision Center. The Agent Inbox, Slack and handing a role to Exterview are groundwork for the agent swarm, which is Upcoming.

​

Added

• AI Employees: four AI Employees, Role Designer, Screener, Coordinator and Report Specialist, work alongside your team, with Smaya for everything else. A reply names the employee that did the work, and work is handed between them under your own permissions. Find them under Settings, Setup, Operations. See AI Employees.
• Groundwork for the Agent Inbox (Upcoming): one queue for what needs a person, where a report to view is answered Proceed, Drop or Talk and a schedule to accept is answered Accept, Decline or Talk. Items reach it once role runs are switched on, which is Upcoming. See Agent Inbox.
• Smaya’s answers stream in, with a panel above the answer showing what the agent did to reach it. See Smaya Assistant.
• Groundwork for asking an AI Employee from Slack (Upcoming), by direct message or mention. It will be read-only. See Smaya.
• Scopes set per client for connected assistants in Claude, ChatGPT and Microsoft 365 Copilot. Upcoming: starting a role run and answering Agent Inbox items from them. See MCP.
• A fourth step in Create Job: the wizard now runs Add Job Details, Add Questionnaire, Add Candidates, then Publish & Outreach. See Jobs.
• View Job Flow: a read-only view of the job’s pipeline, with each stage, its gates and outcomes. Upcoming: rates measured from the job’s own candidates. See Jobs.
• An Agent Role tab on a job, describing the agent’s role and its first message to candidates. See Jobs.
• Pick the interviewer persona for each round and preview the candidate’s experience while creating the job. See Candidate interview experience.
• A video gallery in the Decision Center: switch the score table to interview cards. See Unified report.
• Download candidates as CSV from the candidates toolbar, for users with export permission. See Candidates.
• Panel and HR rounds appear on the calendar at their booked slot.

​

Changed

• Reports is now the Decision Center, in the sidebar and in the address bar. Everything that was in Reports is there. See Unified report.
• Agent Network is renamed Agent Lineage, on the Agent Catalog toolbar. Both are Upcoming. See Agent Library.
• The People Pipeline view on a job reads your real candidate list. See Jobs.

​

Security

• An AI Employee asks before it acts: creating a role, publishing a job and contacting a candidate always need your approval first, and every agent turn runs as you, with your permissions. No AI Employee can set a candidate to hired, rejected or withdrawn. See Governance and control.

​

v1.43.0: 10 September 2026

Candidates are kept up to date over WhatsApp and can move a phone screen themselves, a job’s whole configuration downloads as a document, and the questions written into a job description are no longer discarded.

​

Added

• A candidate is told over WhatsApp when they are shortlisted, when a callback is confirmed, and when they reach the next round. See EX Call.
• A candidate can move their phone screen from the WhatsApp message: picking a new time in the message reschedules the call, without a recruiter arranging it. See EX Call.
• A candidate who misses a call is messaged straight away — the message goes out on the missed call rather than waiting for the next sweep. See EX Call.
• A job’s configuration downloads as a document from the jobs list: it sets out every stage of the interview pipeline in the order it runs, each round’s questions grouped by competency, and carries a DRAFT watermark until the configuration is locked. See Jobs.
• Interview questions written into a job description are carried into the questionnaire: they were discarded, so the second step opened on a set composed from scratch and your own questions appeared nowhere. See Jobs.
• A link to our Microsoft Marketplace listing on the sidebar demo card.

​

Changed

• The screening call follows a rewritten script: the same call, restructured into numbered steps with the language choice asked once at the start, and an explicit exit when the wrong person answers. Screens run before this release used the previous script. See EX Call.
• Screening feedback reads as one paragraph instead of six to eight numbered points, on the Screening Summary and on every AI Justification card. See EX Call.
• Missing skills are reported as one consolidated point — one line covering all of them rather than one per skill, with improvement areas capped at five. Screens analysed before this release were analysed on the previous instruction. See EX Call.

​

Fixed

• AI interview invitations sent over WhatsApp go through again: from 8 September every WhatsApp invite failed to send, because the message was built in a format the number’s approved template does not accept. Invitations sent in that window did not reach the candidate and need re-sending. See Setup communication.
• A resume score’s justification no longer shows the scorer’s raw working: some runs returned a labelled evidence object rather than prose, and it was printed onto the public careers score card as-is. See Careers page.
• The interview report returned to Darwinbox renders reliably — including when it is downloaded rather than pushed. See Darwinbox.
• Creating a job from an uploaded description no longer hangs on the questionnaire step: it sat on “Generating questions for each stage…” with no exit, and while it waited, other requests on the same host slowed down with it. Creating the same job from the prompt was unaffected. See Jobs.
• A candidate’s photo shows on their row in the candidates list: a photo taken from an interview capture appeared on the report but not in the list, which fell back to initials. See Candidates.
• Emails render in light mode in Apple Mail, Gmail and Outlook — they inherited a dark theme that made parts of the message unreadable.

​

v1.42.0: 8 September 2026

Jobs can be favourited and filtered, a role can be changed without re-inviting the person, the careers page is searchable, and a Darwinbox connection now reads your jobs and receives the report as a PDF.

​

Added

• Favourite a job, and filter the jobs list to your favourites: mark the requisitions you are working on and narrow a long list to them. See Jobs.
• Search the careers page by title, type or city. See Careers page.
• A job that runs in several cities lists all of them on the careers page and on the candidate panel. See Careers page.
• Your jobs are read from Darwinbox: an Exterview connection fetches the jobs on your Darwinbox account rather than having them entered a second time. See Darwinbox.
• The interview report is returned to Darwinbox as a PDF, alongside the score and recommendation already written back to the candidate’s record. See Darwinbox.
• Open a candidate’s resume analysis from the Resume AI card on the report. See Unified report.
• Remove several candidates at once while creating a job. See Jobs.
• Select every candidate in the internal pool in one action. See Candidates.
• A report opened from a candidate’s details belongs to the job you opened it from, rather than to whichever job the candidate was last seen on. See Candidates.

​

Changed

• A resume-analysis justification is two to three sentences naming the evidence: the scoring agent was free to write at any length, so justifications ranged from a phrase to several paragraphs. Resumes scored before this release were scored on the previous instruction. See Candidates.
• The screening analysis no longer repeats one observation across feedback, strengths and weaknesses: the same point was being written into each of them in different words. Screens analysed before this release were analysed on the previous instruction. See EX Call.
• The Reports list loads faster: it no longer holds skeleton rows while it assembles an organization-wide view. See Unified report.
• A careers page is published only once a company overview is written, rather than going live with the section empty. See Careers page.

​

Fixed

• A screening call the candidate abandoned is no longer scored: an abandoned call produced a score as though it had been completed. See EX Call.
• A callback the candidate asked for survives a failed capacity debit: the callback time, its confirmation and the next-round step were all skipped when the debit errored, so the candidate was never called back. See EX Call.
• The Scoring Logs export uses the cut you configured: the exported score-delta column was computed against a fixed value of 60 rather than your own threshold, so the file disagreed with the screen. See Settings.

​

Security

• A member’s role can be changed from the users list, instead of removing the person and inviting them again. See Users and roles.

​

v1.41.0: 4 September 2026

Phone screening runs over several rounds, a job’s configuration can be locked once candidates are being scored, and a custom scoring mix no longer moves a score on a stage you did not configure.

​

Added

• Phone screening can run more than one round: a job carries several screening rounds, each triggered, tracked and scored on its own. See EX Call.
• A Phone Rounds tab and a Timeline tab on the expanded candidate row, showing each round’s status, date, duration and score, and what has happened on the candidate so far. See Jobs.
• A job’s configuration can be locked once candidates are being scored: confirming the scoring lock on the first step, or the questionnaire lock on the second, seals the threshold, stage set, question set and weights for that job, and the lock is enforced on the server rather than by hiding the controls. See Jobs.
• A Video view on a job’s candidates, grouped by stage, with each candidate’s interview thumbnail on the card. See Jobs.
• “Nice to Have” is a third skill priority, and Compulsory now reads “Must Have”. See Jobs.
• Choose the voice or avatar that runs each round while creating the job, rather than taking the organization default for every round. See Jobs.
• The Create Job prompt ticks off what it has understood as you type — role, candidate profile, work setup, hiring plan and compensation — without gating the step. See Jobs.
• The public job page lists the rounds a candidate will go through. See Careers page.
• The public job page explains what happens after a resume is uploaded, in four steps, next to the apply card. See Careers page.
• The credits popover shows how far through your subscription term you are. See Billing.
• The dashboard’s Today tab says when there is nothing to show, on both the live rail and the interviews table. See Dashboard.

​

Changed

• An AI interview asks about no more than five skills in one question: a job listing twenty skills had all of them read out in the spoken question, which the interviewer now composes itself. Interviews conducted before this release used the previous instruction. See Candidate interview experience.
• The avatar no longer returns to small talk between technical questions. See Candidate interview experience.

​

Fixed

• A custom AI Screening competency mix no longer moves phone-screen scores: the screening call was scored against the AI Screening mix, so configuring one stage silently reweighted another — dimensions that did not match were dropped and the rest fell back to defaults. Phone screens scored before this release were scored on the mixed weights. See Setup: scoring.
• A name on the transcript that differs from the name on the record is no longer reported as an improvement area on the candidate. See AI interview review.
• Each round runs with the voice or avatar the job chose: the pick was stored on the job and shown back in the form, but every round ran with the organization default. See Jobs.
• Each panel round on a report is its own round, and a round is read only from the job the report belongs to. See Unified report.
• Agent Time on the candidate panel is filled in: it read ”—” for every candidate, including those who had completed both a phone screen and an AI interview. It is the phone screen plus every AI interview round. See Candidates.
• The expanded candidate panel stays in place while the table scrolls sideways, instead of being cut off at both edges. See Jobs.
• The candidate table’s sticky header no longer overlaps the row beneath it. See Jobs.
• Job creation no longer shows a blank questionnaire returned from a cached response. See Jobs.
• A panel round resolves the Teams app registered to your own organization. See Panel round in Teams.
• The camera preview survives being switched off and on again. See Candidate interview experience.
• The interview controls stay on screen on Android when the keyboard opens. See Candidate interview experience.
• A sign-up step no longer stalls before it appears. See Self-onboarding.
• What’s New cards open the tab they describe, rather than a guide.

​

v1.40.0: 1 September 2026

You can now ask questions about a candidate’s report and get an answer from it, voice interviews run on a phone, and the linguistic-integrity check flags fewer honest candidates.

​

Added

• Ask a question about a candidate’s report: a chat opens on the report and answers from that candidate’s own resume, phone screen, voice interview and avatar interview, using the job description and the rounds that were actually run. It answers from the report or says it cannot — nothing carries over between sessions. See Unified report.
• Voice interviews run on a phone: a candidate can complete a voice interview on a mobile browser rather than needing a desktop. See Candidate interview experience.
• A screening call can change language mid-call: EX Call switches language during the call when the candidate does, instead of continuing in the language it started in. See EX Call.
• Report PDFs open with a cover page: a downloaded candidate report now leads with a cover sheet rather than starting mid-content. See Unified report.
• Download an interview transcript as a text file: from the report, without exporting the whole PDF. See Unified report.
• A Documents tab on the candidate details panel, holding what has been collected for that candidate. See Candidates.
• Moving a candidate to another job asks you to confirm it first. See Candidates.
• An AI score you override is kept: the value you set stays on the candidate rather than reverting to what the agent produced. See Candidates.
• Filter candidates on a job by Decision, Score and Notice period, applied across the whole list rather than the page in front of you, and remove several candidates at once. See Jobs.
• Interview thumbnails and recordings on the job’s candidate cards, so an AI interview can be opened from the job without going to the report first. See Jobs.
• A Strong fit count on the Reports list, alongside a rebuilt list that sorts and filters on the server. See Unified report.
• Round scores on the Reports list open the round they came from. See Unified report.
• A customer record now carries an owner, a key account manager, several cities, and a favicon you can upload or pick. See Customers.

​

Changed

• The linguistic-integrity check flags fewer honest candidates: an interview is no longer flagged on a single strong signal, and a candidate who pauses, uses fillers or gives short answers is no longer read as reciting a script. Repeated use of the same template now raises a warning on its own, and natural hesitation can no longer override a flag when the other signals agree. Interviews scored before this release were scored on the previous thresholds. See AI interview review.
• A manual job no longer generates interview questions for you: questions on a manually created job are the ones you write. See Jobs.
• Marking notifications read or dismissed applies immediately, including Mark all as read, rather than after the request returns.
• The post-interview report email has a flatter layout, with the recap card chrome removed and consistent typography. See Setup: communication.
• What’s New now mirrors this changelog directly, so the in-app tab carries the same entries as this page rather than a copy that has to be updated separately.
• Borders are a hairline in dark mode across the product.

​

Fixed

• An unscored phone screen is no longer shown as a score of zero: a screen that was never scored now says so instead of appearing as the worst possible result. See EX Call.
• Repeated feedback points on a voice interview are gone: the analysis was returning the same observation several times in slightly different words. See AI interview review.
• Every report recording has a poster frame, cut from the video itself rather than left blank. See Unified report.
• Report PDFs download with their images intact: images loaded from another domain were being dropped or blocking the export outright. See Unified report.
• The report email uses the candidate’s own preparation capture as its portrait. See Setup: communication.
• The credits usage chart covers your subscription term, not a fixed window that could start before it. See Billing.
• A recruiter whose account is not on Microsoft is told so when scheduling a panel round, instead of the step failing without explanation. See Panel round in Teams.
• The first step of job creation loads its options, and shows the pipeline loading rather than empty. See Jobs.
• A pipeline chosen in the Create Job dialog carries into the manual job flow instead of being asked for again. See Jobs.
• The Resume Score field can be cleared while editing it, and the send-notifications dialog toolbar scrolls. See Jobs.
• A company description can be up to 20,000 characters, and says so when the text is too long instead of failing silently. See Settings.
• “Total Candidates” is labelled correctly on the dashboard, and the KPI bar lines up with the grid beneath it. See Dashboard.
• Start Interview explains why it is unavailable to a candidate who has not been shortlisted. See Careers page.
• Trigger icons, the star toggle and the candidate hover card on a job render as intended. See Jobs.

​

Security

• Every job query and mutation is scoped to your tenant by the same guard: the jobs module now applies one uniform tenant check rather than per-resolver variants, and the job-created event carries a consistent payload. No cross-tenant access was found; this closes the paths by which it could have been introduced. See Users and roles.

​

August 2026

​

v1.39.0: 26 August 2026

Hiring decisions now carry a stated reason, panel and HR rounds are metered against your capacity, and screening runs on rubrics you own.

​

Added

• Recording a decision now asks you why: marking a candidate Recommended or Not a Strong Fit prompts for a reason, and the reason is kept with the decision. See Candidates.
• Decision history on a candidate: see how a candidate’s decision has changed over time rather than only its current value. See Candidates.
• Screening rubrics you own: define your organization’s own rubrics with points-based scoring, and draw the questions asked on a screening call from your own set rather than a fixed one. See Setup: scoring.
• Scoring logs in Settings: a new Scoring Logs section under Operations shows what the scoring agents produced, for the roles in your own workspace. See Settings.
• The report says when a scored answer could not be found in the transcript: rather than showing the score with nothing behind it. See Unified report.
• Your organization’s logo on report PDFs: exported reports carry your logo in the banner. See Unified report.
• Panel and HR rounds show which video platform they run on, including the Teams mark on the agent view. See Panel round in Teams.
• Live notification counts on the Job Created dialog: each channel shows how many invitations have been sent against the total. See Jobs.
• Eligibility tabs on the third step of job creation, and a clearer empty state when nothing matches yet. See Jobs.
• A currency toggle on the credits card, and a clearer state when a free trial has ended. See Billing.

​

Changed

• Panel and HR rounds are now metered against your capacity: they consume from the same capacity ledger as your other rounds. Check your remaining balance before a large panel week. See Billing.
• Interview Activity replaces Cancellation Insights on the dashboard: the card announced on 20 August has been replaced by one that reports interview activity as a whole, with a metric strip and a fuller chart, rather than cancellations alone. See Dashboard.
• Trailer and Highlights are available on the Enterprise plan only: the recording player now leads with Play Recording. On other plans the Trailer and Highlights options appear locked, with an explanation, instead of switching the recording. See Unified report.
• “Not a Good Fit” is now “Not a Strong Fit” wherever a candidate verdict is shown. See Unified report.
• Billing has moved into Setup, under Subscription. See Billing.
• The Enterprise plan card lists the standards that apply, including DPDPA and ISO 27001. Certification status is published on Compliance FAQs. See Billing.

​

Fixed

• Indian mobile numbers beginning 91 are dialled correctly: the country code was being stripped from subscriber numbers that themselves began with 91, so the call never connected. See EX Call.
• A screening question is asked once: a call could re-trigger and repeat a question the candidate had already answered. Calls now leave an activity trail you can read. See EX Call.
• Three phone questions no candidate is ever asked are gone from job creation. See Jobs.
• Continuing a pre-screening no longer asks for an email code: candidates were sent back through a verification step they had already passed. See Careers page.
• Avatar interview recordings play back reliably: some recordings could not be read for conversion, so they never appeared on the report. See AI interview review.
• A scored answer no longer renders blank when the passage it was scored against cannot be located in the transcript. See Unified report.
• The Today tab fills the window: the interviews card and its empty state stopped short of the bottom of the page. See Dashboard.
• Sourced and Applied candidates have their own tabs, and Status is shown once a report exists rather than before. See Candidates.
• One toast per candidate action instead of two.
• The dashboard filter bar scrolls with a fade at the edges rather than overflowing. See Dashboard.
• The credits usage card lays out correctly: capacity grid order, chart height and bar corners. See Billing.
• Report emails render in Outlook dark mode. See Setup: communication.
• Job cards and the create-job flow can be operated from the keyboard, and a candidate’s skills can be revealed without a mouse. See Jobs.

​

Security

• Hiring outcomes can only be set by a person: no automated path can mark a candidate hired, rejected or withdrawn. The consent attestation is enforced before an AI interview can start, and candidates who scored below the bar remain listed rather than being hidden. See Responsible AI.
• Data Visibility is now a second axis of access control: what a user can see is governed separately from what their role lets them do. See Users and roles.

​

v1.38.0: 20 August 2026

Exterview reached general availability on the Microsoft Marketplace, the dashboard reads live figures, and jobs can be published to your careers page in bulk.

​

Added

• The agent inbox now lists one row per report. A candidate who did a phone screen and an AI interview gets a row for each, showing that report’s own score. Previously they had a single row, dated from whichever round won a tie-break. See Dashboard.
• The overview strip shows seven live figures, the cards at the top of the dashboard read current numbers rather than placeholders. See Dashboard.
• Cancellation insight. See why interviews were cancelled, counted across phone, AI and panel rounds. See Dashboard.
• Interview satisfaction on the dashboard, the ratings candidates already give after an interview are now reported back to you. See Dashboard.
• Customer and interview type on recent activity, the Recent tab names the client company and which kind of interview each row refers to. See Dashboard.
• Today counts your working day, not the server’s: the Today tab now takes its day boundary from your organization’s timezone. See Dashboard.
• Publish jobs to your careers page in bulk: select several roles and publish or unpublish them together, with a filter for the ones already active. See Careers page.
• A candidate’s status is now maintained for you. Status is set centrally as a candidate moves through your process, instead of each screen setting it separately. See Candidates.
• Report credits count against your plan’s quota. Agent report credits are drawn from the quota granted on your plan. See Billing.
• Turn off automatic company lookups, the lookup that fills in a client company’s public details can now be switched off. See Customers.

​

Changed

• The daily briefing now arrives Monday, Wednesday and Friday: rather than every weekday morning. See Dashboard.
• A job no longer needs a client company. The customer field is optional when you create a role. See Jobs.

​

Fixed

• Scores are coloured by your decision. A score you had overruled was still coloured against the job’s pass bar, so a candidate you had recommended could show a red score. See Candidates.
• The prompt microphone works again. A quality filter was switching off dictation for everyone.
• Dictation survives a pause, in Chrome, pausing while dictating into an AI prompt field ended the session.
• Recent customers is no longer empty, the widget sorted on a field it could not read, so it returned nothing for every workspace. See Dashboard.
• Report stage tabs count rounds that produced a result. They were counting entries in a list instead. See Unified report.
• The hiring funnel counts real candidates. It read a stage census that was never filled in, so every stage showed zero. See Dashboard.
• A gated figure shows as zero. Where your plan does not include a stage, the funnel showed a dash that read as missing data. See Dashboard.
• Shortlisted and Hired removed from the outcome breakdown. They are pipeline stages, not outcomes, and double-counted there. See Dashboard.
• The avatar column no longer stretches on the candidate grid. See Candidates.

​

v1.37.0: 19 August 2026

Every round is now judged against one of two pass bars you set, and Settings is rebuilt.

​

Changed

• A job now carries exactly two pass bars, both set by you: Resume Score decides résumé screening; Agent Score decides phone, AI interview, panel and HR rounds. A job previously carried four different bars that were applied inconsistently, so a candidate scoring 66 could pass on one screen and fail on another. Every surface now reads the same two. See Jobs.
• The Agent Score picker stores the bar it shows, choosing “70%+” stored 70, not 72. The stored value sat above its own label with no consistent offset, so jobs were rejecting candidates at a bar nobody had chosen. Existing jobs are corrected when they are next read; you do not need to reset anything. See Jobs.
• Settings rebuilt. The settings area has been reorganised around the setup steps.

​

Added

• Star a candidate from anywhere. The star is on the candidate header, the job’s candidate list, the table row, the kanban card and the analysis view, and means the same thing in each. See Candidates.
• Links open the candidate panel where you left it, the panel records which tab is open in the address bar, so a link you send opens the same view. See Candidates.
• The Today view rebuilt: interviews in one table, live cards for what is running now, and a score ring on each. See Dashboard.
• The dashboard pipeline follows your plan. The stages shown are the ones your plan includes. See Dashboard.
• A warning when you pass half your credits, an email at 50% of credit consumption, so a workspace does not discover the limit at the limit. See Billing.
• Résumé screening has its own tab in the report, and HR round scores appear alongside the other rounds. See Unified report.

​

Fixed

• Panel and HR rounds refresh the decision, a candidate who finished a panel or HR round kept the decision from their earlier round on every screen that read it from the server. See Unified report.
• A round’s decision on the report honours your override: the header applied your override to whichever round you clicked, so a round card and the pill above it could disagree on the same screen. Rounds a candidate never reached now stay blank rather than inheriting a verdict. See Unified report.
• Tabs your plan does not include no longer flash before disappearing. See Dashboard.
• Recruiters in an organization with no short name reached a 404.
• Keyboard shortcuts no longer break on an unrecognised key.

​

v1.36.0: 18 August 2026

Screening prompts rewritten, candidate personal data removed from scoring, and a daily briefing.

​

Security

• Candidate personal data is no longer sent to the model that scores them: names, contact details and other identifying fields were present in the scoring prompt. They have been removed, and the scoring parameters moved out of the prompt text. See AI interview review.

​

Changed

• The screening judge has been rewritten: the instructions behind interview scoring, phone screening and the written justification on a report have all been replaced. Scores and justifications produced from today read differently from those produced before it. See AI interview review.
• One scoring calculation across the product: the score on the candidate grid, the report and the dashboard are now produced by the same calculation, and coloured against the same thresholds. See Unified report.
• Emails look the same wherever they come from: invitations, notifications and reports now use one template family. See Setup: communication.
• Bug report severity has fewer options.

​

Added

• A daily briefing at 09:00: a morning email covering what needs your attention, on by default and switchable in settings. See Dashboard.
• A Get Started checklist, a floating checklist and help widget on the dashboard for a new workspace. See Self-onboarding.
• Round scores on hover in the candidate table: hover the AI, Panel or HR column to see that round without opening the candidate. See Candidates.
• An AI Interview score column on the candidate grid. See Candidates.
• Filter a job’s candidates by who they are assigned to. See Jobs.
• Offer and culture blocks in the HR report, drawn from the pre-screening call. See HR round in Teams.
• Candidate photos load reliably. Avatars are served through the product rather than fetched from wherever the résumé pointed. See Candidates.
• Illustrated success and empty screens, screens with nothing on them yet now say so.

​

Fixed

• Transcripts of avatar interviews read in English on the report, the report showed the original language while the rest of the record was translated. See AI interview review.
• A single evaluator is no longer reported as unmarked consensus. One phone screening evaluator was presented as though a panel had agreed. See EX Call.
• Your decision survives a page reload, an override reverted to the AI’s call when the job page was reloaded, while the report header kept showing the override. See Candidates.
• HR round questions come from your organization’s configuration. They were being read from the job instead. See HR round in Teams.
• Long panel interviews produce a complete report. The report budget was sized from the number of questions rather than the length of the transcript, so a long interview was cut short. See Panel round in Teams.
• Long job descriptions produce a complete code assessment, the same sizing problem on the assessment side. See AI code assessment.
• A score meter with no threshold set no longer shows as pending. See Unified report.
• The agent inbox reads its decision from the score rather than from a separate field that could disagree with it. See Dashboard.
• Loading states across candidates, customers and jobs: screens now show the shape of what is loading instead of a blank panel.
• Active and Inactive tabs filter correctly on the candidate grid. See Candidates.
• HR round defaults and staffing. A new HR round could come up without its default questions or its interviewers. See HR round in Teams.
• Keyboard shortcuts on macOS. Shortcuts using special characters did not resolve.
• Assorted dashboard fixes: a sticky View column in the inbox, the wording of the favourite toast, and the height of the figures strip. See Dashboard.

​

v1.35.0: 14 August 2026

HR interview rounds get their own flow, and a batch of reporting and interface fixes.

​

Added

• HR interview rounds. HR rounds are now configured and staffed separately from panel rounds, with their own roster tab, default question set, and a dedicated brief for HR interviewers. See Jobs and Panel round in Teams.
• HR rounds shown in the candidate report, completed HR rounds now appear in the candidate’s report alongside panel rounds, labelled HR. See Unified report.
• Job Home trend view: a new Home tab on the job page shows the hiring funnel over time, with panel and HR feedback combined into a single card. See Jobs.
• Manual free-trial extensions. A lapsed free trial can now be extended by 7 days, repeatably. Extensions show as an “Extended” badge on Settings → Billing, with a notification and an email to the account admin each time. See Billing.
• Subtle interaction motion: buttons, checkboxes, switches, cards, toggles, and other core controls now animate briefly on interaction.

​

Fixed

• AI interview analyzer no longer invents answers, when an interview ended early, the analyzer could fabricate answers for questions the candidate never reached. See AI interview review.
• Panel rounds score correctly. Some panel rounds were not linked to their report and were skipped when scoring.
• Completed rounds always appear in the Reports list, a completed panel or HR round could be missing from the list.
• Candidate detail panel’s Jobs tab redesigned. Shows a job card grid and a decision indicator in the header, and keeps the active tab in view while scrolling.
• Candidate list actions update immediately, deleting or updating a candidate row no longer waits on a refresh to reflect the change.
• Report email shows a real meeting image, instead of a placeholder graphic.
• Scheduling books the round you selected. Previously it could default to the first round in the pipeline instead of the one you picked.
• Reports control bar scrolls instead of overflowing. The stage-pill row is now horizontally scrollable with a fade at the edge.
• HR round with no offer renders correctly. It no longer breaks the page.
• Layout and dark-mode polish: control heights, icon sizing, and dark-mode checkmark rendering fixed across candidates, jobs, and reports.

​

v1.34.0, 13 August 2026

A clearer split between where a candidate is and what you decided about them.

​

Added

• Decision and Status are now two separate things. Status shows where a candidate is in your process. Decision shows the call you made about them. Previously one field carried both, so recording a decision overwrote the candidate’s stage. See Candidates.
• Decision shown on your dashboard, the recent-activity tables now show each candidate’s decision alongside their score. See Dashboard.
• Work mode on a job: mark a role as office, remote, or hybrid. See Jobs.
• More than one location per job, pick several locations for a single role instead of one. See Jobs.
• Panel and HR questions on the job page. See the questions for every round without opening each round. See Jobs.
• Company details filled in for you, when you add a client company, we look up its public details so you do not have to type them. See Customers.
• Choose your typeface, pick the font used across your workspace.

​

Changed

• Jobs get a deadline automatically, new jobs default to a deadline eight weeks out. You can change it. See Jobs.
• The reports list is now a single table. The card view is gone; everything is in one sortable list. See Unified report.
• Candidate summaries are shorter, report summaries and reasons to advance are now kept to a readable length. See Unified report.

​

Fixed

• Leaderboard scores match the report. A candidate could show one score on the leaderboard and a different one on their report. Both now use the same calculation. See Candidate leaderboard.
• Résumé match shown correctly, the report showed how much a skill was weighted rather than how well the candidate matched it.
• Panel reports show only rounds that happened, a scheduled-but-not-yet-held round no longer appears as an empty tab.
• Real names on panel reports, interviewers occasionally appeared as “Speaker 1” instead of their name.

​

Security

• Permission fixes across three roles. Some users could see areas their role was not meant to include.

​

v1.33.0: 11 August 2026

Different questions for each round, and reports you can share outside your workspace.

​

Added

• Each round gets its own questions, every round in your process now has its own set of questions, instead of all rounds drawing from one list. See AI role builder.
• Set how hard a question is, mark questions as easier or harder when building a role. See AI role builder.
• Share a report with someone outside your workspace, send a candidate report to a hiring manager or client who does not have an Exterview login. They confirm their email with a code and get view-only or edit access. See Share report.
• Hiring funnel on your dashboard. See how many candidates are at each stage, and where you are losing them. See Dashboard.
• Assign recruiters to a job. See who owns each role by name. See Jobs.
• Credit balance in the header. Check what you have left without opening billing. See Billing.

​

Changed

• Panel ratings match the job’s skills, interviewers now rate the skills you actually listed on the job, rather than a fixed set that often did not fit the role. See Panel round in Teams.
• Interview recordings show a preview image, recordings open with a frame from the interview instead of a blank thumbnail.
• Audit & Feedback is hidden for now. This section of the candidate report is switched off while we rework how panel feedback is collected. It will return.

​

Fixed

• New jobs always get their questions. Creating a job could finish with no questions, or generate them twice. You now see progress until they are ready.
• Panel interviews for companies using their own Microsoft account, scheduling now works correctly when your organization uses its own Microsoft tenant. See Panel round in Teams.
• Reworded questions still match up, if an interviewer rephrases a question during the interview, the report now lines it up with the original.

​

v1.32.0: 3 August 2026

Job structure, panel interviews in Teams, and a simpler candidate verdict.

​

Added

• Job Architecture Builder: set up your job families, levels, and bands once, then reuse them on every role. See Job architecture.
• Create a job without a JD, start from scratch and watch the progress while we process any document you do upload. See Jobs.
• Interviewer brief posted into the Teams chat, when the assistant joins a panel interview, it drops the brief into the meeting chat. See Panel round in Teams.
• Screening summary in the calendar invite, panel interviewers get the candidate’s screening summary before they walk in. See Panel round in Teams.
• Panellists chosen when the invite goes out: panel rounds follow your pipeline stage, and panellists are picked at send time rather than fixed in advance.
• Search locations by city, type a city and we work out the region. Add a missing location without leaving the job form. See Jobs.
• Better interview calendar: opens on the month view, shows event details on hover, and handles busy days with a popover.
• Drop-off insight on your dashboard. See no-shows and where candidates abandon AI interviews. See Dashboard.

​

Changed

• The candidate verdict is now simply Recommended or Not a Good Fit. This replaces the longer list of verdicts, everywhere in the product. Saved views and exports that grouped by the old verdicts will now show these two. See Unified report.
• Transcripts are saved in English, whatever language the candidate speaks, the saved transcript reads in English. See AI interview review.
• AI voice interviews are English-only, avatar interviews still follow the candidate’s language; voice interviews no longer do. See AI interview review.

​

Fixed

• Panel interviews always produce a transcript. A panel session could finish without one, which left the round with no report. See Panel round in Teams.
• Job creation in Safari, the second step and the side panels could collapse instead of scrolling.

​

Security

• Candidates no longer see their own screening scores. One panel notification included scores in the copy sent to the candidate. It no longer does.

---

​

July 2026

​

v1.31.0: 28 July 2026

Interviews in the candidate’s own language, and much better recordings.

​

Added

• Interviews follow the candidate’s language, the avatar and voice interviewers pick up the language the candidate is speaking and continue in it. See AI interview review.
• Multilingual screening toggle, switch multilingual screening on for your organization. See Set up the AI interviewer.
• Jump to any question in a recording. Recordings are split into chapters, one per question. See Recording chapters.
• Interview calendar. See every scheduled AI interview, drag to reschedule, and handle repeating slots and time zones.
• Set the résumé rubric on a single job, override your organization default for one role. See Set up scoring.
• Trial reminders: nudges on day 7, 14, and 21, with a warning before trial data is removed. See Self-onboarding.
• Ask for access to a report, request access when a report or role has not been shared with you. See Share report.

​

Changed

• All six avatars now speak Indian English, matched to the avatar. See AI interview review.
• Recordings play on the first click, open faster, and no longer download in full before starting.
• The app is set to the light theme while we rework the dark one.

​

Fixed

• Scheduled rounds no longer show a score of zero. A round that had not happened yet could look like a candidate who scored nothing. See Unified report.
• Report PDFs: recording and contact links now work, pages break in sensible places, and the audit log is left out.
• Emails in dark mode, some email apps turned our emails into unreadable colours.

​

Security

• Phone screens no longer mention the salary budget for the role. See EX Call.

​

v1.30.0, 23 July 2026

​

Fixed

• Phone screens are counted correctly. Usage is checked before a call is placed, so you are never billed for a call that could not run. See Billing.

​

Security

• Clearer role permissions. We corrected and expanded what each role can create, view, and edit. See Users and roles.

​

v1.29.0: 22 July 2026

Feature control for administrators, and the candidate leaderboard.

​

Added

• Candidate leaderboard. One ranked view of every candidate for a role. See Candidate leaderboard.
• One overall score across all rounds. A candidate’s score is now the weighted total of every round you have switched on, with a missing round counting as zero.
• Set résumé rubric weights per job, with an explanation of what each weight does. See Set up scoring.
• Publish or unpublish a role to your careers page straight from the jobs list. See Careers page.
• WhatsApp interview invites, send AI interview invitations over WhatsApp. See Setup communication.
• Internal pool finds more candidates. Matching is more forgiving, and close matches are ranked rather than dropped. See EX Match.
• Administrators can turn features on and off per customer, including an emergency off switch.
• Full job details view: questions, agent setup, pipeline, and score breakdown in one place. See Jobs.

​

Changed

• Voice interviews respond faster, the interviewer waits less before replying and keeps answers concise.
• Voice and avatar interviews are billed once per completed report. See Billing.

​

Fixed

• Unanswered questions read “Not Addressed” instead of “Partially”.
• Jobs show their short code (like JOB-1234) rather than a long internal ID.
• Voice interviews no longer go quiet after the candidate’s first answer.
• Your billing page shows your active plan, not the most recently created one.

​

v1.28.0, 16 July 2026

​

Added

• What’s New panel. See the latest updates from inside the app.
• Job details curtain: open a job’s full detail, KPIs, description, and skills without leaving your list. See Jobs.
• Live billing data on the billing page. See Billing.
• Choose which interview types run for each stage of your process. See Setup pipeline.

​

Changed

• Voice and avatar interviews are scored on the role’s competency mix, not the phone screening rubric. Scores for these rounds will read differently from before. See Set up scoring.
• The app now opens in the light theme, with a theme control in your profile settings.

​

v1.27.0: 15 July 2026

​

Added

• Redesigned billing page: credit meter, plan comparison, and per-report pricing in one view. See Billing.
• Completed rounds shown on the candidate curtain. See finished stages without opening the full report. See Candidates.
• Sample answers on uploaded questions, with a badge showing which questions your team added. See Jobs.

​

Changed

• Fit scores are weighted by competency, a candidate’s fit score now reflects how the role weights each competency. See Set up scoring.
• Résumé screening status follows your score thresholds, candidates move on or stop based on the thresholds you set. See Set up scoring.

​

v1.26.0, 14 July 2026

​

Added

• Bulk upload questions from a spreadsheet. See Jobs.
• Locked phone screening questions in the job creation wizard, so every candidate for a role is asked the same things.
• Callback rescheduling: if a candidate cannot talk, we capture a better time and call back automatically. See EX Call.
• Duplicate a job to reuse its setup for a similar role. See Jobs.
• Word document résumés are now accepted alongside PDFs.
• Candidates are re-scored automatically when you change a job, so your ranking stays current.
• CC on candidate scheduling emails, always on. See Setup communication.

​

Security

• Clearer magic-link errors. You can now tell an expired link from an invalid one, and see how long links last.

​

v1.25.2, 13 July 2026 (hotfix)

​

Fixed

• Report PDFs are generated on our servers, which makes them consistent regardless of your browser.
• Candidate counts across your whole workspace now exclude duplicates.

​

v1.25.1, 10 July 2026 (hotfix)

​

Fixed

• A follow-up fix to the 10 July release.

​

v1.25.0: 10 July 2026

Your careers page, and multi-round AI interviews.

​

Added

• Public careers pages: a branded page listing your open roles, with job details, an apply flow, and a pre-screening step. See Careers page.
• Candidates can apply and be scored automatically, a résumé uploaded through your careers page is scored and shortlisted without anyone touching it.
• Multiple rounds of AI interviews, schedule and track more than one AI round per candidate. See AI interview review.
• Internal talent pool: upload résumés once, then find and rank those candidates for any future role. See EX Match.
• A reason is required when you change a candidate’s stage, so your pipeline history explains itself.
• Phone screening reports show voice as its own stage.

​

Changed

• Phone screening scores come from your configured rubric. Scores that could previously be estimated are no longer produced at all. A missing score now reads as missing. See Set up scoring.

​

v1.24.1, 9 July 2026 (hotfix)

​

Fixed

• A follow-up fix to the 8 July release.

​

v1.24.0: 8 July 2026

​

Added

• Job detail and apply pages on your careers site, running on live data. See Careers page.
• Print and full-report modes for candidate reports, with an improved PDF layout. See Unified report.

​

Changed

• The AI interviewer follows one governed set of instructions for both voice and avatar interviews, so interviews are consistent between candidates. See AI interview review.

​

v1.23.1, 8 July 2026 (hotfix)

​

Added

• Careers landing page at your own address. See Careers page.
• Applications through your careers page are scored and shortlisted automatically.

​

Changed

• Reports are only generated once an interview is complete enough to be worth reading, and you are not billed for reports we do not produce. See Setup gates.

​

v1.23.0: 7 July 2026

​

Added

• Scoring weights in the job wizard: set and save how much each part of your rubric counts, with validation so the weights add up. See Set up scoring.
• Question bank in job creation, plus a rich-text job description and a careers URL in settings.
• No duplicate interview invites. Candidates who have already completed a round are skipped unless you deliberately re-invite them.
• Bulk question import from CSV or Excel.

​

Security

• Sign-in protection: a CAPTCHA appears after three failed attempts, repeated attempts are rate-limited, and password reset no longer reveals whether an email address exists.

​

v1.22.0, 2 July 2026

​

Added

• Better validation on customer records, clearer errors when creating or editing a client company. See Customers.
• New interviewer voice replacing one of the earlier options. See AI interview review.

​

v1.21.0, 1 July 2026

​

Fixed

• Small corrections following the 30 June release.

---

​

June 2026

​

v1.20.0: 30 June 2026

Sharing, scoring, and a faster bulk upload.

​

Added

• Redesigned report sharing. See who has access at a glance, and share with your whole workspace in one step. See Share report.
• Send an interview invite straight from a candidate record, and copy their contact details in one click. See Candidates.
• Filter candidates by customer, export to CSV, and download résumés. See Candidates.
• Generate a sample answer for any question you have written. See Jobs.
• Type your own values for job skills instead of picking only from the list.
• Departments come from your organization settings everywhere they appear. See Setup org structure.

​

Changed

• Résumé scoring is faster and cheaper: uploading the same file twice reuses the first result instead of scoring it again, and large batches are processed in parallel. Scores are unchanged. See EX Match.

​

Security

• Sign-in hardening: accounts lock after repeated failed attempts, password resets are rate-limited with a CAPTCHA, and sign-in no longer reveals whether an email is registered.
• Signing in on one tab now signs you in across tabs.

​

v1.19.0, 26 June 2026

​

Security

• Tightened network access to our API, so it is reachable only through our protected front door.

​

v1.18.0, 25 June 2026

​

Changed

• Groundwork only, no change to how the product behaves.

​

v1.17.0, 24 June 2026

​

Added

• Play phone screening recordings directly in the candidate record, without downloading them. See Candidates.
• Bulk résumé upload shows live progress and processes files in parallel, so large batches finish much faster. See EX Match.
• Voice and avatar interviews ask your own questions, the questions you set in the job wizard are used and rotated across candidates. See AI interview review.
• Edit an existing job through the same wizard used to create one, with the job code shown in the breadcrumb.

​

v1.16.0: 23 June 2026

​

Added

• Internal talent pool in the Add Candidates dialog: pull in matching candidates you already have, ranked, and import them in bulk. See EX Match.
• Set and save a candidate verdict during job creation.
• Empty states across the dashboard, so a new workspace explains itself instead of showing blank cards. See Dashboard.
• Six interviewer voices, four international and two Indian. See AI interview review.

​

Changed

• Updated plans and pricing. See Billing.

​

Security

• Compliance guides published for SOC 2, ISO 27001, ISO 42001, and India’s DPDP Act. See Setup compliance.

​

v1.15.0, 19 June 2026

​

Added

• Clearer error screens, a page that fails now explains what happened instead of going blank.

​

v1.14.0: 18 June 2026

​

Added

• Candidate activity timeline, with an audit log on every report, so you can see who did what and when. See Candidates.
• Star a candidate to mark the ones you want to come back to.
• Agent Report column on the jobs view, with a phone column that adapts to context.
• Confirmation messages on every action in jobs, phone screening, and questions, so you always know whether something saved.

​

Security

• Full role-based access control across the product. What each person can see and do now follows their role everywhere, front and back. See Users and roles.

​

v1.13.0, 15 June 2026

​

Added

• Role-based report sharing. Choose what a person can see when you share a candidate report. See Share report.

​

Fixed

• Clearer candidate tables: better spacing, and score tooltips that read in a sensible order.

​

v1.12.0: 12 June 2026

​

Changed

• Updated pricing: résumé screening at $2 and phone screening at$5 per report, with a $2,000 free trial. See Billing.

---

​

May 2026

​

v1.11.0. 29 May 2026

​

Added

• Set your interview voice and avatar during onboarding, and have those choices carry into your settings automatically. See Self-onboarding.
• Job locations come from your organization settings, with salary labels that follow the location’s currency. See Jobs.
• Compliance trust bar during onboarding, so you can see our compliance posture and current certification status as you set up.

​

Fixed

• Phone screens that got stuck are now retried automatically.
• Sign-in errors read in plain language instead of showing a technical message.
• Signing out returns you to a clean sign-in page.

​

Security

• Fixed administrator sign-in for our support team.

​

v1.10.0. 28 May 2026

The first release we tagged, carrying the ten days of work that led up to it. Setup became live and configurable.

​

Added

• Settings → Setup is wired to live data: Users, Roles, Pipeline, Config, Scoring, and Skills all read and save for real, instead of showing placeholder values. See Settings.
• Pipeline templates, start from a ready-made hiring pipeline instead of building each one from scratch. See Setup pipeline.
• Invite your team by email: invitations go out as a sign-in link, and you can resend one that expired. See Setup users.
• Pick the customer and location while creating a job. See Jobs.

​

Changed

• Your saved scoring weights are now actually applied: the résumé rubric weights and stage weights you set drive résumé scoring, the leaderboard, and the unified report. Before this, they were saved but not used. See Set up scoring.
• System pipelines are marked read-only, so it is clear which ones you can edit.

---

​

Before we tagged releases

Everything below predates 28 May 2026, when we started tagging every release. The work is real and is in the product. But we were not recording exact release dates yet, so these entries are dated by when the work landed, not by a recorded deployment. Versions below v1.10.0 are numbered from that history.

​

v1.9.0. 17 May 2026

​

Added

• Setup → Skills. See the skills each AI agent is configured to assess. See Agent skills.
• Panel and HR rounds trigger themselves from the hiring pipeline, instead of being scheduled by hand.
• Teams interviews record automatically when scheduled through Exterview. See Panel round in Teams.

​

Fixed

• Forgot-password flow is clearer, and a mistyped URL now lands on a proper page instead of a blank one.

​

v1.8.0 (16 May 2026 Internal work only) no change to how the product behaves.

​

v1.7.0. 15 May 2026

​

Fixed

• AI interview audio no longer plays twice.

​

v1.6.0. 14 May 2026

​

Added

• Pipeline templates in Setup. See Setup pipeline.
• Manage your company locations and departments from organization settings. See Setup org structure.
• Redesigned organization profile step during sign-up. See Self-onboarding.

​

v1.5.0. 13 May 2026

​

Added

• Upload your logo and favicon during onboarding, and crop them in place. See Self-onboarding.
• Phone numbers accept an international dialling code wherever you enter one.
• Your sign-up progress is kept if you step backwards through the wizard.

​

v1.4.0. 12 May 2026

​

Added

• Early groundwork for a browser-based panel room. The room itself, Advanced Video Room, is upcoming and is not available yet.
• Crop your logo and favicon when uploading them.

​

v1.3.0. 11 May 2026

​

Added

• Panel interview screen: video stage, panellist strip, and a side rail with the candidate overview, agent notes, and suggested questions. See Panel round in Teams.

​

v1.2.0. 10 May 2026

​

Added

• Schedule a panel round from the job, with the assistant joining the meeting. See Panel round in Teams.
• Sign in with email and password alongside Microsoft sign-in. See Setup roles and access.

​

v1.1.0 (9 May 2026 Internal work only) no change to how the product behaves.

​

v1.0.0. 8 May 2026

​

Added

• Share a candidate report by link: a public reader with view-only, restricted, and internal access levels, and a share button on every report. See Share report.

​

Changed

• Rebuilt our emails so they render consistently across mail clients. See Setup communication.

​

v0.5.0. 29 April to 7 May 2026

​

Added

• Panel interview assistant: transcription, analysis, and the panel report pipeline that produces a report after a panel round. See Panel round in Teams.

​

v0.4.0: 22 to 28 April 2026

​

Added

• Settings: profile, company, setup, and activity logs in one place. See Settings.
• Recruiter dashboard and analytics. See Dashboard.
• Multiple regions and languages, with locale-aware formatting. See Setup org structure.
• Score justifications on hover. See why a candidate scored the way they did, on phone, voice, and avatar rounds. See Unified report.
• Candidate report fills in from the résumé, and keeps what it extracted. See Unified report.
• Trigger a phone screen from the job page, with pipeline-aware columns and deadline notes. See EX Call.

​

v0.3.0: 15 to 21 April 2026

​

Added

• Avatar interviews, end to end: scheduling, the interview itself, and a consolidated report. See AI interview review.
• Candidate leaderboard with a podium view, stage selection, and recruiter-set stage weights. See Candidate leaderboard.
• Play back a voice interview recording from the report. See Unified report.
• Report rounds follow the job’s hiring pipeline, so the report matches the process you configured. See Setup pipeline.
• Analytics dashboard. See Dashboard.

​

v0.2.0: 8 to 14 April 2026

​

Added

• Hiring pipelines: create, edit, and delete the stages a candidate moves through. See Setup pipeline.
• Questions shown stage by stage during job creation, so you can see what is asked at each round. See AI role builder.
• Upload candidates while creating a job. See Jobs.
• Role-based access settings, and About and Company tabs in settings. See Users and roles.

​

v0.1.0, 1 to 7 April 2026

​

Added

• AI phone screening, outbound calls with an AI screener, reported back against the role. See EX Call.
• Bulk résumé upload into a job. See EX Match.
• Candidate report sections: education, experience, job description match, company, and an audit trail, with AI justifications you can expand. See Unified report.
• A default hiring pipeline is created for you when your organization is set up. See Setup pipeline.
• Jobs and job details run on live data. See Jobs.

---

​

FAQs

​

Related