Shipping every day.

Resume-to-profile comparison signals now remain visible even when the values agree.

iCIMS job setup now supports reset substages for remote review actions.

Bulk review tasks now have a longer execution window, reducing timeouts on large candidate groups.

Identity mismatch findings are now grouped under Application Analysis.

Ashby connections through Kombo now support advancing and rejecting applicants from Endorsed.

Ashby connections through Kombo now import the core records needed for Applicant Review.

Resume signal thresholds now distinguish cautionary and high-risk patterns more consistently.

Fraud actions now use clearer decision labels, reducing ambiguity when reviewers mark or clear an outcome.

The Fraud Risk filter now remains enabled in entitled Applicant Review workflows instead of disappearing unexpectedly.

Fraud Detection now consistently describes its output as predicted risk.

A missing Lever resume attachment no longer blocks the rest of an application from loading.

The ATS jobs loader now retries transient provider failures more reliably.

Missing Greenhouse attachments no longer fail the broader candidate sync.

Greenhouse screening-question data now comes from validated job-post data included in the normal sync.

The Applied Within filter now offers more useful recent-date ranges for incoming applications.

Candidate stage tabs now use distinct icons for unreviewed, holding, advanced, and rejected pools.

An oversized Workday attachment no longer stops the candidate's other available attachments from syncing.

Applicant Review now handles missing, malformed, and provider-specific ATS links without breaking the candidate view.

Bulk Advance is now available by default whenever the connected ATS supports remote stage changes.

iCIMS jobs can be matched using remote requisition IDs, improving linked-job synchronization.

Jobs deleted in a connected ATS are now recognized during sync, keeping the Endorsed jobs list aligned.

Applicant counts now follow the same rejection-stage visibility rules as the candidate list.

Lever candidate records now link directly to the corresponding ATS application.

Department-triggered fraud checks are queued independently from automatic evaluation.

The candidate grid now includes location, current role, employer, and education.

Configured iCIMS connections can write a candidate's Fraud Risk score and Endorsed profile link back to the ATS.

Eligible Applicant Review accounts can include a Fraud Risk range in Bulk Reject.

ATS stage changes now tolerate a deleted rejection reason whose name is no longer available.

iCIMS sync now skips an application with no current stage instead of failing the entire application batch.

Legacy fraud signals no longer appear in aggregate summaries, reducing duplicate or confusing evidence.

Fraud signals now show clearer hierarchy and supporting detail, making individual findings easier to interpret.

Fraud and review actions now target the correct application when a candidate appears in multiple review contexts.

Applicant Review combines applied-date, seen-status, source, and score controls in a clearer filter experience.

Resetting a rejected Greenhouse applicant now unrejects the record before moving it back to the configured review stage.

Kombo sync now accepts applications whose current stage is missing and skips only the incomplete application.

iCIMS connections through Kombo now support the core data needed for Applicant Review.

Recruiters can filter the jobs list by creation date, making large workspaces easier to narrow.

Workday connections through Kombo now accept more real-world job, candidate, contact, and attachment data without failing the sync.

Applicant Review can detect when a Greenhouse application was deleted remotely and synchronize that state.

Lever now identifies stale rejection reasons during stage changes and prompts recruiters to choose an active reason.

Greenhouse rejection stages are now offered only when the connected integration supports them.

Candidate links back to Greenhouse are preserved when ATS metadata is normalized.

Bulk Reject remains available when no Fraud Risk filter is active, restoring the expected Applicant Review action.

Fraud checks can use available enrichment records even when no personal email was found.

Greenhouse rejection reasons are verified and synchronized against the ATS, including reasons deleted remotely.

A missing Greenhouse rejection reason now produces a specific, recoverable message during a stage change.

Applicant Review briefly reuses its request-verification token, reducing avoidable interruptions across repeated page actions.

Fraud checks now parse an available resume before detection when stored parsed content is missing.

Fraud analytics now reports medium- and high-risk percentages and reliably renders its charts.

Fraud analytics once again shows trends over time and includes unchecked candidates in the default population.

Greenhouse stage changes now use more spacing and an additional retry, reducing failures from provider rate limits.

Fraud Detection can apply an account-wide usage allowance while preserving more specific user allowances.

Reviewers can apply a configured fraud rejection reason when marking a candidate as fraud.

Clearing a fraud reason now works on the first click, so the saved candidate label immediately matches the reviewer's choice.

Resetting a review decision now sends the corresponding remote stage change when the ATS supports it.

Candidate fraud checks now reuse one enrichment request per candidate, avoiding duplicate lookups and inconsistent results.

Reviewers can filter for unseen candidates, with profile views recorded so the queue reflects what has actually been opened.

Eligible Applicant Review accounts now show live fraud-check progress and hide the indicator when processing is complete.

Accounts without Fraud Risk filtering now see a clear locked state and next step instead of a missing control.

Attachment conversion now uses request-scoped queues, preventing one batch of scanned resumes from interfering with another.

Fraud-enabled Applicant Review accounts can filter candidates by a range of Fraud Risk levels.

Resume parsing deployments now include standard PDF fonts, improving extraction from files that rely on embedded font mappings.

Verify Candidate now handles a replacement file cleanly, so reviewers can retry with the correct resume.

The latest fraud scoring model now runs by default for all supported checks, without a rollout flag.

Complex fraud checks can run for up to five minutes, reducing timeouts before all signals finish.

Fraud Detection now shows resume parsing as a distinct progress step, so reviewers can see what the check is waiting on.

All-numeric LinkedIn profile identifiers are now rejected as invalid, preventing bad links from weakening identity checks.

Education and experience mismatch tooltips now explain what differs between a resume and profile.

Resume-related fraud signals are grouped into clearer subcategories, making the evidence easier to scan.

Verify Candidate now waits for resume parsing to finish before a check can be submitted, preventing incomplete requests.

Applicant Review uses a softer score palette, making dense candidate grids easier to scan.

Applicant Review now calls its ranking metric Criteria Fit, making the score's meaning clearer in filters.

The candidate grid now labels integrity results as Fraud Risk, clarifying what the score represents.

Fraud evidence is available from the organization-wide candidate profile for candidates imported from a connected ATS.

Fraud checks now carry extracted resume text into detection, giving resume-based signals the evidence they need.

The candidate grid now shows each available Fraud Risk level, so reviewers can triage risk without opening every profile.

Resume uploads use a unified parsing endpoint with more precise errors, so failed files are easier to recover and retry.

Fraud checks now return a clear error when no indicators are available, preventing an empty result from being mistaken for a completed assessment.

Introduced Important Signals, top-signal summaries, and access to the complete evaluated evidence set.

Moved threshold filtering into the shared backend and expanded important-signal display.

Added jobs, recruiters, rejection reasons, phone data, and attachment support.

Added a new version, corrected indicator pagination, and recalculates when the latest score is missing.

Corrected failures in extension resume processing.

Introduced separate verified-identity and fraud evidence treatments.

Surfaced duplicate resume evidence in the candidate view.

Repaired create, update, verification, and deletion behavior.

Review controls now update the candidate’s fraud-label record.

Added aggregate fraud analytics, report UI, and CSV export.

Extracted and displayed relevant links from candidate profile data and submitted resumes.

Introduced a unified risk meter, signal display, and fraud actions beside the candidate profile.

Connected the fraud action directly to Applicant Review rejection.

Treats non-public LinkedIn responses as a valid unavailable state.

Fixed skipped, indeterminate, and vendor-error cases.

Shows zero counts when the unreviewed stage is not configured.

Added default checks for VoIP and phone risk across multiple data sources.

Added invalid-number evidence, dynamic explanations, and dependent signal execution.

Added a dedicated product section and path into the experience.

Recent prior rejections can bypass a second Greenhouse email.

Added rejected-for-fraud totals to the applicant funnel.

Supporting evidence can now be surfaced from the biography section.

Safely handles resume text that previously produced an invalid regular expression.

One unavailable attachment no longer blocks the entire sync.

Later provider syncs retrigger ingestion for jobs that did not finish.

Corrected duplicate matching in action creation.

Highlights the supporting subphrase instead of only a broad profile section.

Job and role identifiers now use stable distinct keys.

Repaired a client/server rendering mismatch.

The configured description now persists with the job.

Recruiters can reject without sending the configured email.

Fixed a selection crash in nested job lists.

Adjusted retry timing and attempts.

Malformed fetch responses no longer break the review queue.

Allocates the processing route enough memory to avoid failures.

Permits useful probabilistic language while retaining validation.

Estimates recruiter time saved by reviewed rejections.

The filters page no longer crashes.

Setup stops correctly after the intended candidate set.

Unified feedback processing and upgraded the tuning path.

Repaired criteria-weight updates that were lost after tuning.

Detects and reconciles missing files across major integrations.

Later syncs retrieve candidate files that were absent from the original import.

Identical review jobs cannot be started twice.

Feedback controls, agent responses, and loading states became part of setup.

Added role and job views for stage distribution, hires, and rejections.

Increased timeouts and retries for slow integration endpoints.

Attachment ordering uses the provider’s creation time.

Repaired stale rejection-reason selection.

Retriggers initial ingestion when a retried sync succeeds.

Clarified rejection-email delays and stage movement.

Rewrote the action copy around its consequences.

Prompts users to reconfigure when an ATS stage is no longer valid.

Inactive stages are hidden from decision controls.

Stage changes now handle applications deleted in the connected ATS without leaving stale review records behind.

Candidate exports now include criterion results and related fields.

Corrected sidebar and selector behavior reported by users.

Highlighted the referral badge on application profiles.

Evaluation now honors text already extracted from an uploaded file.

Tuned timeouts for application fetches and rejections.

Handles accounts whose access role is missing.

Related jobs and review pages now navigate through their parent role.

The queue now respects each applicant’s latest review action.

Saving no longer remounts the form with an empty state.

Improved filtering, score calculation, and completion across related jobs.

One failed criterion no longer hides every other criterion.

Fixed failures when adding many templates to a partially configured job.

Reviewer decisions no longer unexpectedly revert.

Improved the core setup and reviewer UI.

Fixed unintended auto-nomination and duplicate evaluation runs.

Repeated save clicks no longer create duplicate nominations.

Added filter editing, counts, validation, and save-state awareness to the review sheet.

Removed a race with saved browser state.

Added multi-select, range filters, and live filtered counts.

Transcribes page by page and handles empty or refused results.

Added longer timeouts, retries, and backoff.

Criteria fields no longer steal focus while a recruiter types.

Repaired the Applicant Review order-by filter.

Fixed the rejection decision control so the selected reason is applied correctly.

Criteria now open in compact expandable cards that keep prompt text, labels, weights, and template settings together.

New setup pages guide recruiters through ATS configuration and job settings with clearer save status and next steps.

Applicant Review recommends criterion changes based on reviewer feedback and evaluation patterns.

Recruiters can open a focused editing sheet, update or infer criteria, save with keyboard shortcuts, and return directly to candidate review.

Applicant Review adds a persistent sidebar across dashboard areas for clearer navigation.

Recruiters can add common role criteria from guided templates, including company-size and industry matches populated from organization context.

Applicant Review converts secondary and legacy compound-document files into PDFs so recruiters can view them in the candidate record.

Applicant Review flags subjective or potentially biased criteria as recruiters type and suggests clearer alternatives.

Applicant Review accepts both response shapes Ashby returns for select fields, preventing valid screening answers from being dropped.

Selected teams can choose which job stage feeds the unreviewed queue; changing it starts evaluations for candidates in that stage.

Candidate review now shows files tied to the current application while retaining candidate-level files that apply across roles.

Applicant Review now carries all application custom-question responses into a consistent screening record for evaluation.

Applicant Review can include an application’s source when evaluating source-specific job criteria.

Long application-source labels stay within the candidate profile header instead of crowding contact actions.

Candidate review now separates profile context and supporting files into clear tabs, with richer work-history and company details.

Applicant Review now uses application screening answers alongside resumes when evaluating candidates against job criteria.

Recruiters can select a Greenhouse rejection reason while rejecting candidates from Applicant Review.

Recruiters can include rejected or archived-stage applicants when reviewing a job.

Applicant Review can ingest jobs, candidates, and application state from BambooHR through the connected ATS pipeline.

Attachment imports are throttled in controlled batches to prevent race conditions during conversion and processing.

Applicant Review now turns resume attachments into structured candidate data before generating evaluation summaries.

Teams can receive recurring email summaries for jobs with evaluated candidates still waiting for review, with direct links back to the job.

Bulk review pagination now retrieves every queued candidate action, including the item previously missed at a page boundary.

Bulk rejection actions now use the rejection email template selected for the job.

Stage-change workflows now resolve the acting ATS user from synchronized account data instead of depending on a live remote lookup.

Job score histograms now exclude candidates already rejected from the unreviewed pool.

Empty values in Greenhouse multi-select responses are filtered before ingestion, keeping screening data clean and processable.

The Jobs List adds search, filtering, sorting, pagination, and direct links to ATS and external job records.

Recruiters can archive inactive jobs from the production Jobs List.

Applicant Review now handles Greenhouse's global Hired stage even when the ATS omits a job association.

Greenhouse rejection email templates now accept the identifiers returned by the ATS and display the saved choice correctly in job settings.

Greenhouse Silo customers can complete the connection and authorization flow directly during onboarding.

Applicant Review shows how much of a job’s candidate evaluation work has completed.

Ashby and Greenhouse screening questions follow the correct ingestion path for each provider.

Recruiters can reject a filtered group of matching applicants in one Applicant Review action.

Recruiters can narrow an applicant pool by whether candidates meet a specific review criterion.

Applicant Review uses one scroll area instead of overlapping scrollbars on longer candidate lists.

Candidate profiles show the source associated with an application beside the reviewer’s other candidate context.

Applicant Review keyboard actions keep working as recruiters move between candidates and filters.

Applicant Review can schedule a selected Greenhouse rejection email template with a recruiter’s reject action.

Advance and reject actions in Applicant Review update the corresponding application stage in Greenhouse.

Recruiters can assign different weights to job criteria and update the resulting candidate ranking.

Recruiters can focus the review queue on applicants within a selected relevance-score range.

Advance and reject actions in Applicant Review update the corresponding stage in Ashby.

Recruiters can configure up to ten criteria per job, while platform accounts can configure up to fifteen.

Recruiters can choose whether Applicant Review should infer criteria from the available job context.

The Endorsed public footer links directly to the existing Trust Center.

Platform teams can query candidate records through a dedicated public API endpoint.

Platform teams can retrieve ranking and supporting summary data for a single candidate.

Recruiters can add or edit job criteria after an Applicant Review run has started.

Platform teams can list, create, retrieve, and delete role-specific Applicant Review criteria through authenticated endpoints.

Candidate-specific ranking responses include the applicant’s relevance score for the requested job.

Applicant Review reorders the visible queue as new candidate evaluations complete.

Recruiters can work through candidate evidence and review controls in a clearer review layout.

Recruiters can configure role-specific review criteria directly in the Applicant Review job workflow.

Applicant Review distinguishes evidence that cannot support a confident yes-or-no evaluation instead of forcing a guess.

Candidate ranking responses now include remote candidate and application identifiers from the connected ATS.

The existing ranking lookup now uses a candidate-centered API route name.

Platform teams can retrieve candidate attachments for Applicant Review through the public API.

Platform teams can create and retrieve Applicant Review jobs and candidate records through authenticated endpoints.

Applicant Review now imports interview stages from connected ATS data and maps each application to the right point in the hiring process.

Recruiters can sign in, ingest candidate resumes, and review ranked applicants in a connected workflow.

ATS connection now stays in a focused onboarding layout and sends recruiters directly into setup after a successful connection.