Candidate rediscovery software (2026): three different machines wear the same label

Search "candidate rediscovery software" and you get a tidy promise. Your database is full of people you already paid to attract, the software finds the ones who fit your new role, you place someone this week instead of sourcing for three.

The promise is real. The category is not. "Rediscovery" is a word sitting on top of at least three genuinely different machines, sold at prices that differ by a factor of thirty, solving three problems that only sound the same. Buy the wrong machine and you get exactly what a lot of agencies got: a feature you switched on, a list of people you had already placed, and a quiet decision never to open that tab again.

This page is the sorting job. What each machine actually does, how to tell which failure you have, what each option costs with prices read off vendor pages this month, and where each one stops working.

Contents

• Quick summary
• Why should you listen to us?
• The three machines
• Diagnose your own failure first
• Machine 1: search over the record
• Machine 2: refresh from outside
• Machine 3: a read of the document
• The comparison table
• Is rediscovery even a category?
• When rediscovery software is the wrong purchase
• Where Gemsift fits, honestly
• FAQ

Quick summary

• "Candidate rediscovery" describes three machines. One searches the parsed record your ATS built when the resume came in. One refreshes and extends that record from external data. One re-reads the original document against the new role and tells you why someone fits. Vendors use the same word for all three.
• Your failure mode tells you which to buy. If the person you missed was in the results and you scrolled past, that is an ordering problem. If they were never returned at all, that is a reading problem. If you found them and could not reach them, that is a freshness problem. Three different products.
• The loudest rediscovery marketing is aimed over your head. Phenom and Avature both put talent rediscovery in their own CRM descriptions, and along with Beamery and Eightfold they are enterprise talent platforms sold through a sales cycle. A five person agency is not their buyer.
• Prices split sharply. As of August 2026: Manatal from $15 per user per month on annual billing, Crelate from $85, Bullhorn Starter at $99, Recruiterflow Platform at $149, SeekOut Recruit Core at $149 per month for three seats, Loxo Basic at $169, hireEZ from $494 per month for a solo recruiter. Several vendors publish a price for the filing cabinet and quote separately for the part that reads it.
• The mechanics question that settles it: when you open a new role, does the tool match against fields it extracted two years ago, or does it read the document again now? Most public docs do not answer this. Ask in the demo and watch what happens.
• Gemsift is machine three. It is the AI-native ATS you move into with one export file, and it re-reads your whole base against each new role with a plain English reason per pick. Free, $99, $299 and from $899 a month. It is not a sourcing tool and not an enterprise compliance platform, and we say where it does not fit below.

Why should you listen to us?

I build one of these. Gemsift is an AI-native ATS for recruitment agencies, so treat this page as informed and interested rather than neutral, and check the prices yourself before you sign anything.

What makes it worth reading anyway: the prices below were read off vendor pricing pages during the week this was published, not copied from a listicle that copied a listicle. Where a vendor does not publish a number, this page says so instead of quoting a range from a third party marketplace. Where public documentation does not say how a product matches, this page says that too, which turns out to be most of them.

The framing comes from about forty conversations with agency and in-house recruiters over the past two months. One of them, an agency owner, told me his old system made it "easy to resurface" candidates because they were "housed by position", which is a precise description of the trap this whole page is about. Another said the candidates she wished she had not passed on were real, but "these are not from our ATS, more from sourcing and revisiting them". Both are the same lesson from opposite ends: the word rediscovery hides where the loss actually happened.

The three machines

Machine 1: search over the record

When a resume arrives, your ATS parses it into a record. Title, employer, dates, a skills list, some tags, an indexed blob of text. From that moment on, most "AI matching" is a query against that record. Semantic search is a better query than boolean, but it is still a query, and it is still against the extraction rather than the document.

This machine is fast, cheap, and already in your stack. Its failure is structural, not accidental: it can only return what the extraction captured. The operations manager who ran a warehouse and would be excellent in logistics recruitment does not come back for "logistics" if the parser did not write that word into her record.

Machine 2: refresh from outside

The second machine solves a different problem. Your record for someone is from 2023. Since then they changed employers, changed titles, and changed email addresses. Enrichment and sourcing platforms reconnect that record to current external data, often against a large external profile graph, and present the result as rediscovery.

This is genuinely valuable and it is not the same job. It answers "is this person still here and still reachable". It does not answer "is this person right for this role". Two of the products in this space also generate duplicates in your base, because the external profile they find and the record you already hold are not always recognised as the same human.

Machine 3: a read of the document

The third machine goes back to the original resume, reads it against the specific role in front of you, and returns an ordered shortlist with a stated reason per person. The unit of work is the document and the role together, not the record and the query.

This is the newest of the three and the hardest to verify from a website, because every vendor in categories one and two now describes their product in language that sounds like this one. The distinguishing evidence is usually in the help documentation rather than the marketing page, and it shows up in two details: whether the run is triggered by a human clicking a button on one job, and whether the output contains a reason you could paste into a client email or just a percentage.

Diagnose your own failure first

Before you look at a single vendor, work out which machine you are missing. Take a role you filled recently the hard way, and find one person in your base who would have been a legitimate candidate.

Was that person in the search results, and you scrolled past? Your problem is ordering and attention, not retrieval. A better ranking with reasons attached fixes this. Machine three.

Was that person never returned by any search you ran? Your problem is reading. The record does not contain the words that would have surfaced them, so no query over the record will find them, no matter how semantic. Machine three, and a search upgrade will waste your money.

Did you find them, and then discover the phone number was dead and they had moved twice? Your problem is freshness. Machine two.

Did you find them, reach them, and they said no because you last spoke two years ago and never followed up? No software fixes that. That is a relationship gap and it belongs in your own reactivation process, not in a purchase.

Most agencies I speak to discover they have the second problem and were shopping for the first. That is the expensive mistake this page exists to prevent.

Machine 1: search over the record

Bullhorn. The dominant agency ATS. Bullhorn publishes prices for its small agency plans, at $99 per user per month for Starter and $165 for Core, with Pro and Max quote based. Notably, Amplify Search and Match is listed as its own item under AI and automation pricing rather than being folded into those plans, so the published number is the system of record and the matching is a separate line. Verify current figures on their pricing page.

JobAdder. No published prices at all. The page states plainly that instead of one size fits all pricing they build a tailored proposal, with agency plans banded by user count (1 to 5, 6 to 20, 21 plus). JobAdder's search and match is delivered in partnership with Daxtra, whose technology is semantic search over parsed and indexed resume data. That is a mature, well understood machine one, and Daxtra also powers matching inside other products, which is worth knowing when two vendors demo suspiciously similar results.

Recruit CRM. Three plans, Pro, Business and Enterprise, with a monthly and annual toggle and up to 20 percent off annual. We could not read a dollar figure off the pricing page when we checked in August 2026, so get the number from them directly rather than from an aggregator.

Manatal. The cheapest credible option here by a wide margin: $15, $35 and $55 per user per month on annual billing for Professional, Enterprise and Enterprise Plus, or $19, $39 and $59 monthly. Automated recommendation of candidates based on a job description is shown as included across all three paid tiers, which is unusual. If your base is small and your budget is real, this is the sensible place to find out whether match over the record is enough for you.

Machine 2: refresh from outside

SeekOut. Recruit Core is published at $149 per month paid annually, which the page states as $1,788 per year, or $179 billed monthly, and it includes three seats. Larger sourcing and full funnel tiers are custom. For a small agency this is one of the few enterprise flavoured tools with an honest entry price.

hireEZ. Solo recruiters start at $494 per month with a seven day trial. For larger teams the page is explicit that pricing is configured against the stack hireEZ replaces rather than a flat per seat rate, which is a fair description of a sales led model. hireEZ grew out of the sourcing world, and that is still where its centre of gravity is.

Gem. Gem names an AI Talent Rediscovery Agent that resurfaces past candidates across your CRM and ATS, which is about as explicit as this category gets. Pricing is mostly custom and tied to your company headcount, with a startup program for 1 to 10 FTE shown at $130 per month billed yearly against a $270 list price. Gem is built for in-house talent teams more than contingency agencies, and the seat model reflects that.

Loxo. A hybrid, and honest about it. Loxo publishes a free tier for a single user and Basic at $169 per user per month, then puts Loxo Source, its 850 million plus profile talent graph, and all of its AI agents in the Professional tier, which is quote only. So the published price gets you an ATS and CRM, and the rediscovery-shaped capability sits behind a conversation. That is the pattern we flagged in AI-native ATS for recruitment agencies: the price you can see is often not the price of the thing you came for.

Machine 3: a read of the document

Recruiterflow. AIRA Matchmaker is the clearest documented example outside our own product. Recruiterflow's help documentation describes opening the job, clicking Generate, setting criteria in plain language rather than boolean, and getting a criteria score you can click into per candidate, with the guidance to always review the reason for the match. It also states the model does not consider demographic attributes. Pricing: the Platform plan is $149 per user per month and explicitly does not include AIRA, and the AIRA plan is custom. Note the trigger: it is a button on a job, not something that fires on its own when a role opens.

Crelate. Essentials is $85 per user per month paid annually for up to two users and 20,000 contacts, and Business at $119 adds the AI Assistant. Business Plus, with the advanced AI assistant, is custom. Crelate is a strong agency ATS and the AI tier boundary is at least visible, which is more than most.

Skima AI. Positions explicitly on explainable matching, which is the right axis. Its pricing page did not publish figures when we checked in August 2026, and third party listings disagree with each other about what it costs, quoting both $49 and $79 per user per month. Treat both as unverified and ask the vendor.

Gemsift. Ours. Details in the section below rather than buried in a list where you cannot see the bias.

The enterprise platforms in this conversation, Phenom and Avature (both of which describe talent rediscovery inside their CRM), plus Beamery and Eightfold, are real products doing real work at organisations with thousands of employees. They are sold through a sales process, implemented over months, and priced accordingly. If you are a 1 to 20 person agency, they are not your shortlist, and any listicle that ranks them for you has not thought about who is reading.

The comparison table

Prices as of August 2026, read from each vendor's own pricing page. They change, and several vendors run promotions, so confirm on the site before you buy. "Quote" means the vendor does not publish a number for that tier.

The last column is the one that costs you money quietly. A rediscovery feature that waits for a human to remember it exists gets used enthusiastically for two weeks and then never again. That is not a knock on any specific vendor, it is how buttons work.

Is rediscovery even a category?

Short answer: no, and that is useful to know before you go shopping.

There is no meaningful analyst category or review site category called candidate rediscovery software. These products live in applicant tracking systems, recruiting CRM, talent intelligence and sourcing categories on G2 and Capterra. The phrase had its moment in the late 2010s, when a handful of startups built businesses on the pitch that your ATS was a graveyard and AI could resurrect the silver medallists. Those companies were acquired or absorbed, and the capability moved inside the systems of record, where it now shows up as AI matching, similar candidates, or an agent with a name.

The practical consequence for a buyer: you are not choosing between rediscovery products. You are choosing an ATS, a CRM, or a sourcing layer, and evaluating how good its rediscovery is. That reframes the decision. Migration cost, seat pricing and daily workflow matter more than the feature you came for, because you will live in the system every day and use the rediscovery on the days you open a role.

It also means the vendor claims are unusually hard to check. Numbers like "70 percent of candidates in an ATS are never contacted twice" and "rediscovered candidates convert three times faster" circulate widely in this space, and they trace back to vendor marketing rather than to any audited study. They may well be directionally right. Do not put them in your business case as facts.

When rediscovery software is the wrong purchase

• Your base has no resumes attached. If a migration from an old system brought over names, emails and notes but the documents were lost, no machine three product can read what is not there, and machine one has very little to index. Fix the corpus or accept the ceiling.
• You place fewer than a handful of people a month from a small base. A few hundred candidates and two open roles is a reading job you can do yourself in an afternoon. Software here is a subscription against a problem you do not have.
• Every role you work is genuinely novel. Rediscovery pays off on repeat and adjacent roles. If no two briefs in your year resemble each other, your base is a weaker asset than your network.
• Your candidates are all placed or unavailable and your statuses do not say so. Every tool in this list will cheerfully surface people you placed last quarter if your status fields do not mark them. That is a five minute conversation with a vendor and a genuine deal breaker if the answer is vague.
• Your real bottleneck is outreach capacity, not selection. Then you want engagement automation, not matching. Different aisle.

Where Gemsift fits, honestly

Gemsift is the AI-native ATS an agency moves into when its current ATS has become a filing cabinet. You import with one export file, and when you open a role it re-reads your entire candidate base against that role and returns a client-ready shortlist with a plain English reason per person, not a percentage. On the three machines above it is squarely machine three, the re-read fires when the role opens rather than waiting for someone to click, and it is in every plan including the free one rather than behind a quote.

Prices as of August 2026, and confirm them on the site since they change: Free at $0 for a base of 100 people with no card, Solo at $99 a month for 1,000, Agency at $299 a month for 5,000 with three seats, and Scale from $899 for larger bases. You can run it against built-in sample data without uploading a single real resume, which matters if client confidentiality means you cannot trial tools on live candidate data.

Where it is not the answer:

• You need heavy integrations, job board posting, or enterprise compliance tooling. Keep a traditional ATS. This is not that.
• You are a pure outbound headhunter. Your problem is finding people who never applied to you. That is sourcing, and hireEZ or SeekOut are built for it.
• Your base is tiny or your volume per role is single digits. Read them yourself.
• You want relationship history back. We read what is in the documents. We cannot recover a conversation nobody wrote down.

Start free, no card

FAQ

How is this different from the AI matching already in my ATS?

Usually two things: when it fires, and what it hands you. Most ATS matching waits for a person to press a button on one specific role, and returns a ranked list with a score. The questions worth asking your current vendor are whether the re-read happens on every new role without anyone remembering to trigger it, and whether you get a reason you can read and overrule rather than a number you have to trust. If the reasoning is not visible, you cannot exercise the meaningful human review that regulators increasingly expect of automated candidate ranking.

My database is already searchable. Why would I need this?

Because searchable and surfaced are different things. Boolean and semantic search both return people who match the query you wrote against the record your parser built. Position based resurfacing returns people whose old title resembles the new one. Both work well for obvious re-fits. The candidate worth finding in an old base is usually the one whose substance fits and whose vocabulary does not. A useful self test: when did your search last surprise you with someone you would have missed? The mechanism behind that gap is in why keyword filters keep rejecting good candidates.

Do I have to clean up my database first?

Not for the reading. A tool that needs your data pre-organised to be useful is telling you it depends on tags and field hygiene rather than on the documents, which is worth knowing before you buy it. What you do need before any outreach is a suppression list: already placed, do not contact, unsubscribed, duplicates. That is not cleaning for the software's benefit, it is protecting your list.

Half the resumes in my base are AI-written now. Won't AI just be fooled by them?

That is the problem this is built for. The right approach looks at what someone actually did rather than how polished the writing is, cross checks claims against the work history, and flags generic filler for your review. An AI-written resume is not a weak candidate, so nobody should be disqualified for looking AI. The point is that the strong person who wrote plainly does not sink to the bottom.

I already do this in Claude or ChatGPT. Why would I pay for a system?

If you already paste role context into a model for each brief, you have proved the point: that context is what makes the call good. What you would pay for is the part you rebuild by hand every time. Structure, so you are not reassembling prompts and files per role. Consistency, so the same bar applies across thousands of records and across your whole team. A base that remembers, so the next role does not start from zero. If a tuned setup already gives you all three, keep it.

Is my candidate data safe?

You can evaluate Gemsift on built-in sample data without uploading a real resume, which is the honest answer for anyone under client confidentiality rules. If your agency has data governance requirements, ask for a data handling summary covering where data lives and how long it is kept before you upload anything, and export or delete whenever you want.

What should I ask on a rediscovery demo?

Four questions, in this order. When I open a new job, does the match run on its own or do I click something? Are you matching against fields you extracted when the resume arrived, or reading the document again now? Show me the output for a candidate: is there a reason a client could read? And does it respect placed, unavailable and do-not-contact statuses, or will it hand me people I placed last quarter? Vague answers to the second and fourth are the ones that cost you.

---

Related reading

• AI-native ATS for recruitment agencies for the three questions that separate a real AI-native system from a badge, tier by tier.
• How to reactivate your candidate database without burning it for the process once the software has surfaced the people.
• The shift to AI-native recruiting for why the bottleneck moved from finding people to judging them.
• Why keyword filters keep rejecting good candidates for the mechanism behind the buried candidate.
• Best AI resume screening tools for recruitment agencies for the wider tool landscape and its prices.