Gem's AI rediscovery scores and explains past candidates automatically the moment a recruiter searches for talent, no separate tool required. Fraud detection for those candidates, though, is not live yet. Some users report duplicate profiles the system misses, and nothing records what a hiring manager decided about a rediscovered name.
This guide covers what Gem's rediscovery does today, then where connecting external AI tool enhances verification, deeper duplicate resolution, and an audit trail that Gem does not yet provide.
What Is Talent Rediscovery?
Talent rediscovery means searching and matching candidates already stored in your company's database against newly opened roles. It covers applicants from closed jobs, candidates parked in a talent network, and anyone who advanced partway through a hiring process before the role got filled or shelved.
A year or two of application history turns into an active sourcing channel once those stored candidates get checked against every new opening automatically. Most ATS platforms leave that to a recruiter's memory. Rediscovery performs the check every time, without anyone needing to recall a name.
Does Gem Provide Candidate Rediscovery?
Partially, Gem's AI Rediscovery, part of its AI sourcing agent, surfaces past candidates from a company's ATS and CRM who match a recruiter's search criteria, combined with new external prospects. Prospect Search, an older, filter-based tool, covers similar ground manually across six supported ATS platforms.
However, neither tool verifies what it surfaces. Gem announced an Applicant Fraud Detection Agent in December 2025, still unreleased as of this guide, so a rediscovered candidate's claimed skills and resume content reach a recruiter exactly as submitted.
Nothing scopes a search by application age, tracks a hiring manager's decision on a name, or triggers automatically when a new requisition opens without a recruiter starting a search.
How to Rediscover Candidates in Gem
Gem's AI Rediscovery already finds and scores past candidates. Connecting Skima AI does not repeat that step; it verifies what Gem surfaces, resolves duplicates that Gem's own detection misses, and keeps a record of every hiring manager decision on a rediscovered name. Below is a 5-step workflow for adding that AI tool on top of what Gem already does:
Step 1: The Native Path, Start With AI Rediscovery
When a recruiter searches for talent through Gem's AI Sourcing Agent, AI Rediscovery pulls in matching past candidates from the connected ATS and CRM alongside new external prospects, each with a match score. Plus, visible history like past applications and interview notes.
This step surfaces candidates from a company's own history with a score and reasoning already attached. However, it does not verify what it surfaces, resolve duplicate profiles beyond a narrow set of matching fields, or track what a hiring manager decides about any name. Skima AI addresses those three gaps directly, without repeating Gem's own scoring.
Step 2: Connect Your Gem Account With Skima AI
Create an API key from Team Settings in Gem, with read access on Candidates, Applications, Attachments, and Jobs, and write access on Notes, Custom fields, and Tags. This connection uses Gem's own REST API directly, with no middleware required.
- Read access: candidate profiles, application data, job details, and resume attachments, including the identifiers (email, phone, LinkedIn URL) used later for duplicate matching and the resume text used later for verification
- Write access: candidate custom fields, tags, and activity notes
- Pulled data: candidate contact details, resumes, application stage and source, job title and job description for scoring, plus prior work history and project details for verification context
- Event triggers: candidate created or updated, application submitted, stage changed, and attachment added
This is the point where the two identifiers that later steps depend on get pulled in: the contact fields duplicate matching runs on, and the resume content verification checks against.
Step 3: Configure Custom Fields and Enable Webhooks
Add the candidate custom fields Skima AI writes to: Match Score, Reasons, and Shortlist Link, or let Skima AI create them automatically. Reasons is also where verification outcomes surface. A flagged inconsistency, say a skill claim that doesn't match the candidate's stated project history, reads as part of that same text instead of staying hidden.
Enable webhooks for candidate, application, and attachment events so Skima AI provides the endpoint and secret. If webhooks aren't available, Skima AI falls back to a 10 to 15 minute polling interval instead.
Before rolling this out further, test the connection on two or three candidates. Upload a resume and confirm the Match Score and Reasons appear, with a verification note included if one applies. Then submit sample hiring manager feedback and check that it logs to the activity note correctly.
Step 4: Scan, Deduplicate, and Verify Against a Live Requisition
Open the job inside Skima AI, from the Jobs list or from within the job itself. Click the three-dot Actions menu and select Rediscover Candidates.
A modal opens with options to narrow the scan: select a specific candidate segment, or leave it open to every synced Gem record. Set the candidate created date range: All time, Last 6 months, Last 1 year, or a custom window. Click Apply.
The scan then works through three passes before any candidate reaches the results list:
- Duplicate check first: candidates get matched on email, phone, or LinkedIn URL, then on name, location, and company together as a backup signal. Anything flagged does not block the scan, it routes to the Duplicates tab for the recruiter to resolve separately, so the ranked list that follows isn't built on top of unresolved duplicate records.
- Verification second: each candidate's claimed skills get checked against their stated project history and job duties, and the resume gets checked for signs of AI-generated content. A flag from this step appears in that candidate's Reasons text rather than being scored as if the claim were confirmed.
- Scoring last: only after those two passes does the requisition's match score and ranking get calculated, so the number a recruiter sees already reflects a checked, deduplicated candidate rather than a raw match.
A "Performing Rediscovery" notification appears while this works in the background, and turns green with a refresh option once it completes.
Step 5: Review, Filter, and Sync Feedback Back to Gem
Refresh the page and open the Rediscovered source tab inside View Candidates. Each match carries a Skima AI score and reason bullets, tagged with a Rediscovery source label. Filter by Industry, Work Mode, Experience Level, Notice Period, Location, and Salary Range to narrow the list.
Once the shortlist looks right, generate a secure, no-login link and send it to the hiring manager. They click thumbs up, thumbs down, or maybe on each name, with an optional comment. Skima AI writes that decision back into Gem as an activity note on the candidate record, creating the audit trail Gem's own rediscovery does not produce on its own.
What If Your Gem Database Is Messy, Duplicated, or Stale
Gem identifies potential duplicate profiles under specific conditions: matching first name, last name, and primary email, or a primary email listed in another profile's emails, paired with a matching company name.
When these criteria align, a notification appears at the sidebar's top, but only while a recruiter views one of the matching profiles. Potential duplicates are not displayed as a list, and merging is limited to one candidate at a time. Some users, in verified reviews, say the system often fails to catch duplicate entries reliably.
On the ATS side, Gem automatically merges applications from the same candidate applying for the same job. It also merges applications sharing a LinkedIn profile URL across different jobs. This LinkedIn check successfully identifies a common scenario: a candidate applying for multiple roles. However, it misses candidates who applied without a LinkedIn URL or those whose profiles lack that specific field.
Conversely, Skima AI approaches this as ongoing maintenance rather than chance detection. It prioritises matches based on email, phone, or LinkedIn URL, then checks name, location, and company for additional confirmation. When a match is found, a comparison modal displays differences in Name, Title, Company, Location, Email, Phone, and LinkedIn, tagging each mismatch as "Differs."
Recruiters can choose which values to retain and merge records, or add the profile as new if the two genuinely differ. Multiple flagged duplicates can be resolved simultaneously, with each case logged under the Duplicates tab, categorised into In Database, From Uploads, and Resolved sub-tabs for future reference.
5 Benefits of Candidate Rediscovery with Skima AI in Gem
Connecting Skima AI to Gem's existing sourcing and discovery tools adds 5 specific strengths a recruiting team does not get from either native tool alone:
- Verified Candidates, Not Just Scored: Every rediscovered match gets checked against project history and resume-authenticity signals before it reaches a shortlist, a check Gem does not yet offer.
- Duplicate Resolution Beyond Email Matching: Records get matched on phone and LinkedIn URL too, with bulk resolution instead of one profile at a time, catching what narrower native detection misses.
- A Decision Trail for Every Rediscovered Name: Hiring manager responses log automatically against the candidate record. In a documented 30-day pilot, this feedback loop lifted hiring manager satisfaction by 24 points on NPS.
- Faster Time-to-Screen: Rediscovered candidates arrive already matched, verified, and deduplicated, cutting time-to-screen by 58% in that same pilot compared to reviewing unverified matches one by one.
- Interviews Double From the Shortlist: The same pilot saw interviews booked from the shortlist double compared to the prior process.
5 Best Practices to Rediscover Talent in Gem
Getting real ROI out of this connection means using each tool for what it is actually built to do. The 5 practices below cover verification, duplicate hygiene, and fitting the added layer around a team already sourcing through Gem daily:
- Let Gem Source and Skima AI Verify: Use AI Rediscovery for the initial search and score. Route the results through Skima AI's scan before outreach, so nothing gets contacted before its claims are checked against resume evidence.
- Automate the Full-Database Scan on Job Creation: Skima AI's setting under Preferences and General starts a database scan the instant a new job gets created inside the connected account, rather than waiting for a recruiter to start a search.
- Extend Reverse Search Beyond One Job: Prospect Search returns candidates for one job at a time based on manual filters. Reverse Search starts from one of those candidates and returns every open job they match, ranked by score.
- Segment Scans by Candidate Created Date: Choose a 6-month or 1-year window when a role calls for someone recently active. Widen to all time for a niche or hard-to-fill role where the applicant pool has stayed thin for years.
- Review the Duplicates Tab After Every Import: Gem's own import process can create duplicates when a name or email does not match exactly across files. Checking the From Uploads sub-tab after each import catches what a mismatched header or typo lets through.
Is Candidate Rediscovery Worth It for Your Recruitment Team?
Gem's AI Rediscovery already gives a company a working search across its own history. So the question is whether unverified matches, narrow duplicate detection, and no audit trail on hiring manager decisions create enough risk to justify a second layer.
For a company filling sensitive or compliance-heavy roles, where a misrepresented resume or an unlogged rejection reason carries real consequences, verification and an audit trail matter regardless of database size. However, for a small team with a few low-stakes hires a year, that risk is smaller, and the added subscription cost is harder to justify.
Pilot Candidate Rediscovery in Gem Now
For a company with a multi-year Gem history, a pilot with Skima AI can deliver high ROI. Start with two or three live requisitions, ideally roles the company has filled more than once before, so there is a real historical pool to verify and clean up.
Connect Skima AI to Gem, configure the custom fields, and test the connection on a few candidates before going live. Track how many rediscovered candidates get flagged during verification, how many duplicate resolutions happen in bulk instead of one at a time, and how many hiring manager decisions get logged that would have gone untracked before.
Skima AI's documented 30-day pilot with Gem also reported a 58% drop in time-to-screen and interviews from the shortlist doubling. From there, expanding to more requisitions becomes a decision backed by the pilot's own numbers.
Frequently Asked Questions
1. Does Gem have a built-in talent rediscovery feature?
Partially, Gem's AI Rediscovery automatically surfaces past ATS and CRM candidates during a search, with a score and stated reasoning. It does not verify what it surfaces, resolve duplicates beyond a narrow set of fields, or track a hiring manager's decision on a rediscovered name.
2. Does Gem verify candidate claims or detect resume fraud?
Not yet, Gem announced an Applicant Fraud Detection Agent in December 2025. It has not shipped as part of AI Sourcing or Rediscovery as of this guide.
3. How do I add verification and duplicate control to Gem's rediscovery?
Connect Skima AI through Gem's API, configure custom fields for Match Score and Reasons, then start a scoped scan on a live requisition. The scan checks for duplicates and verifies claimed skills against resume evidence before any candidate reaches the shortlist.
4. What are the benefits of connecting Skima AI to Gem?
Rediscovered candidates arrive already verified and deduplicated, cutting time-to-screen by 58% and doubling interviews from the top-10 shortlist in a documented 30-day pilot. Every hiring manager decision logs automatically, and satisfaction rose 24 points on NPS.
5. Is Skima AI safe to integrate with Gem for talent rediscovery?
Yes, Skima AI scores and verifies candidates using resume-based evidence only, with no automatic rejections, and logs every hiring manager decision and shortlist action for audit. Data stays encrypted in transit and at rest, backed by a signed Data Processing Agreement.
