Ashby checks each applicant against your criteria, and your team makes the call

Ashby is an applicant tracking system, the software that holds your job applications. Its AI application review checks each applicant against criteria you set. Its Talent Rediscovery searches past candidates for a new role. Application review is on every plan. Talent Rediscovery is on Plus and Enterprise. Both draw on AI credits.

Who should buy Ashby, who should not, and the price

In one line

Ashby checks each applicant who has not opted out against the criteria you set for a role. It shows the reasoning. A second tool searches past candidates for a new role. Your team decides who moves forward.

Buy it if

You want AI sorting inside the system that holds your applications. You want to see the reason behind each result. Your company can carry the legal duties, because Ashby’s terms place them on you.

Do not buy it if

You need AI that moves applicants on or rejects them without a person. Ashby’s review tool leaves that step to your reviewers. Workable Agent can do that once someone switches it on. You need Talent Rediscovery on the entry plan. It is on Plus and Enterprise only.

Price

Foundations costs $400 a month for companies of up to 100 employees. It is set in five tiers by workforce size. Plus suits 101 to 1,000 employees and Enterprise suits 1,000 or more. Ask Ashby for the price at your headcount. Checking one candidate uses one AI credit.

How Ashby sorts applicants and finds past candidates

You write the criteria for each role. Ashby can suggest them from the job posting, and you can reuse criteria from other roles or add your own. An administrator switches the feature on. Editing a job’s criteria needs a higher permission. Ashby then reads each CV against each criterion. When there is no CV, it uses the education and work history on the candidate’s profile.

Each criterion gets one of three results. They are meets, does not meet and undecided. A column shows the percentage of criteria each candidate met. Ashby calls it the AI Job Criteria Met Percentage. Sorting by it moves the closest matches to the top of the review queue. The AI result does not change the Average Score, which holds your human reviewers’ ratings. A first evaluation takes from a few seconds to a few minutes.

A person then decides. When reviewing applicants in a batch, the reviewer advances or rejects each candidate and must pick a rejection reason. Ashby also has auto-reject rules, which are separate from the AI. They reject people whose application-form answers match conditions you set. They run only when the application is submitted.

Talent Rediscovery works on the people already in your Ashby database. It matches them against a new job’s criteria and sorts them into groups by how many criteria they meet. One group is Warm Lead. These are people who showed interest but never interviewed. Ashby builds signals from application history, interview ratings, email threads and AI labelling. By default it leaves out people with certain negative signals. Each person comes with prior feedback, application history and refreshed work history, location and education where available. Since 31 July 2026, job-linked Rediscovery opens to an All Candidates group. The percentage met is a default filter there. Your first scan of past candidates can take up to 24 hours.

Rediscovery can also start a message sequence to the people it finds. A user must accept or reject the AI-written text before the sequence goes out. Ashby’s AI Interviewer is a separate voice interview product in private beta, so judge it on its own.

What works in your favour

Every result comes with its reasoning

You can click any result to see why the AI gave it. Ashby also links to the evidence behind each result. A reviewer can flag a result as incorrect or misleading. Ashby tells reviewers to read the full application next to the AI result before they trust it.

Your criteria are checked for problems before use

Ashby checks each criterion for clarity and for wording that could discriminate under US equal opportunity rules. It flags what it finds. You can still go ahead with a flagged criterion, but Ashby recommends that you change or remove it.

Your applicants’ data is not used to train AI

Ashby redacts personal details from CVs before it sends them to an AI model. Neither Ashby nor its AI suppliers train models on your data. The suppliers are OpenAI, Amazon Bedrock and Google Gemini. They process your data only to answer your request, and they do not keep it after the session.

Candidates can opt out, and the AI skips them

Applicants see a Learn More link under the Submit Application button. It shows the notice and lets them opt out of automated processing. If they opt out, Ashby skips the AI check. You can write the message candidates see. Since 28 May 2026 you can also report on who opted out.

The review tool is on every plan

Ashby released AI-Assisted Application Review to all customers on 10 September 2024. It is available on every Ashby plan. You pay for it through AI credits, Ashby’s prepaid usage units.

What the bias audit, Ashby’s customers and Ashby itself report

An outside audit found no group below the 80% line

The audit firm FairNow audited the AI that checks CVs against your job criteria. It did this in August 2024. It looked at race, gender and their combinations. No group had a selection rate below 80% of the most favoured group’s. One applicant-criterion is one candidate checked against one criterion. The audit covered 123,610 of them across the US and Europe, from 1 January to 12 August 2024. Results marked uncertain, 11.4% of the total, were left out.

The audit leaves one legal question open

FairNow did not judge one question. That question is whether the model is an automated employment decision tool under New York City Local Law 144. Ashby had the audit done even though it does not treat its AI as one of those tools. Whether the law applies to you is a question for your own legal advice.

Ashby customer January found 15 best matches in more than 500 applications

January used the review tool to find its 15 best matches among more than 500 applications. Ashby customer Dave uses the tool to handle large numbers of applications. Dave’s Head of Talent says it fits their plan of setting success criteria first. NETGEAR’s Director of Talent Acquisition calls it a huge time saver.

Ashby’s VP of Talent reviewed 1,500 applications in six hours

The tool has processed millions of CVs. Jim Miller, Ashby’s VP of Talent, reviewed 1,500 applications in six hours with it. More than half of Ashby’s customers use Ashby AI every day.

Where Ashby stops, from AI mistakes to candidate notices

Ashby is not responsible for what the AI gets wrong

Under Ashby’s AI terms, it is not responsible for AI output or its accuracy. You must check the output yourself. You are solely responsible for what you put in and for making sure it follows the law. Ashby’s customer terms cap what it owes you. The cap is $500 or what you paid in the twelve months before the claim, whichever is greater.

Telling candidates about the AI is your job

Under the EU AI Act, your company is the one using the AI. Under the EU’s data protection law (GDPR), your company is also the one in charge of candidates’ data. The AI-use notice is your responsibility, in your own policy and wording. Some candidates may never have seen the notice. These include people added by referral, by hand, through the Chrome extension, or possibly through Indeed Apply. A setting hides AI results for anyone who has not seen it.

AI checks pause when your credits run out

When all AI credits are used, new AI evaluations pause until you add more. Unused credits do not roll over to the next contract year. Ashby warns you in the app when you reach 80% of your credits.

Form questions are only checked for an answer

Criteria built from application-form questions check whether the candidate answered. They do not judge the answer. Resumes carry the judgement, or the education and work history on the profile when there is no CV.

Health and government data need care

Ashby cannot sign the contract that HIPAA, the US health privacy law, requires before a supplier handles health data. You must not share data covered by HIPAA with Ashby. Ashby does not have a FedRAMP Authorization to Operate, the US government’s security approval for cloud services.

What Ashby costs and what the contract says

Ashby sells three plans. Foundations costs $400 a month for companies of up to 100 employees. Ashby prices Foundations by total workforce size, in five tiers. Plus is for 101 to 1,000 employees. Enterprise is for 1,000 or more. Price depends on company size, usage and commitment. An annual commitment earns a 10% discount.

Ashby bills for each user given a paid seat. Other user types are free. For example, 20 paid seats at $795 per seat a year would cost $15,900 a year. Treat that as an example, not a price list. Not every account is on seat pricing.

Most Ashby AI features are free. The ones that run on large volumes of data use AI credits. One credit pays for one candidate checked against up to 50 criteria. Rediscovery uses one credit for each candidate returned, up to 250 credits for one search. Fraud checks use two credits for each candidate. They flag applications with possible fraud signals from the device, internet address, email address and phone number.

Foundations includes 1,500 credits a month, or 18,000 a year, for the whole account. Plus includes 2,500 a year for each paid seat. Enterprise includes 12,500 a year for each paid seat. Annual and multi-year plans pay $500 a year for 5,000 more credits. Month-to-month Foundations buys 1,000 for $100 a month. Say a team only checks applicants. A role with 500 applicants uses 500 credits, so three such roles in one month use the whole Foundations allowance. Ten paid seats on Plus give 25,000 credits a year. Ashby suggests estimating need by multiplying past applicants per role by the hires you plan this year.

The contract runs for the term on your order or one year. It renews for one-year periods unless either side opts out at least 30 days before. Month-to-month plans carry a 30-day money-back guarantee. Invoices are available only on annual plans. Ashby can change the terms by posting a new version on its website. A change takes effect 30 days later unless you reject it in writing. Ashby may remove features at any time. If a change cuts a large part of your order’s features, you can end that order. You must give 30 days’ written notice.

The terms promise only to try reasonably hard to stay available all day, every day. A separate service agreement sets a 99.9% uptime target for each quarter. It applies only if your contract refers to it. The credit is 10% of quarterly fees for uptime from 99.9% down to 99.0%, and 25% below 99.0%. Disputes go to the courts in San Francisco, California. Ashby owns the usage data it collects and may use it to improve its products. It may also use your name and logo in its sales and marketing.

Leaving takes planning. Reports export to a spreadsheet file (CSV). A single candidate’s export can include the profile, social links, work and education history, emails and notes. Bulk CV export is not possible in the app, so Ashby’s support helps. A data export asked for at cancellation is handled by hand by Ashby’s team. Ashby may permanently erase your data once an account has been unpaid, suspended or ended for 30 days. A deleted candidate is erased after 10 days, and backups after 30 days.

You could also find and screen people with agents built on your own systems. We have mapped what that looks like. The HR Agent Map

Eight questions to put to Ashby

  1. Which plan and how many paid seats do we need? What is the total with the AI credits we expect to use? Good answer: A written quote that lists seats, included credits and the price of extra credits.
  2. What happens to a live hiring round when our AI credits run out? Good answer: A written top-up price and a stated time to add credits.
  3. Does your bias audit cover the percentage column recruiters sort by, and is there a report newer than August 2024? Good answer: A report that covers the sorted view, with results for each group.
  4. Will you write the 99.9% uptime target and its service credits into our order form? Good answer: Yes, in the order form, so the separate service agreement is part of our contract.
  5. Can changes to your terms need our written agreement instead of taking effect 30 days after you post them? Good answer: Yes, or at least notice sent to a named contact on our side.
  6. How does every candidate who arrives by referral, by hand or through the Chrome extension see the AI notice? Good answer: Every route shows the notice, or the AI result stays hidden until the candidate has seen it.
  7. How do we get all CVs and AI results out when we leave, and when do you delete them? Good answer: A named file format, a delivery date and a written deletion date.
  8. Which AI services read our applicants’ CVs today, and where do they process them? Good answer: A named service and region for each, with notice before either changes.

Alternatives to Ashby

The alternatives are Greenhouse, Workable, Gem, SeekOut and Juicebox. Buying nothing is the sixth option.

Greenhouse, Workable and Gem also hold your applications. SeekOut and Juicebox search for people outside your database.

Greenhouse sorts applicants into five match groups. They are Strong, Good, Partial, Limited and Needs manual review. Greenhouse uses only data from the CV. It never advances or rejects a candidate. Its Past candidates tab filters earlier applicants and does not use AI match scores.

Workable’s AI finds qualified past candidates in your talent pool when you post a new job. Standard costs $299 a month, Premier $599 and Enterprise $719, all with annual payment. Workable launched Workable Agent on 13 March 2026 to source, screen and shortlist. One setting automatically disqualifies candidates who miss criteria marked Disqualify. Another moves candidates to a stage you choose when score thresholds and must-have conditions are met.

Gem scores and ranks every applicant against your criteria. Gem holds applications in its own applicant tracking system. Its AI rediscovery agent searches past candidates in its candidate database and in that system. The AI makes no decision to advance or reject. A knockout question rule can reject a candidate automatically when you set it up.

SeekOut and Juicebox both connect to Ashby. SeekOut can export profiles into Ashby, and it can search your Ashby candidate data. Juicebox searches people outside your database. It can import a job from Ashby, save profiles into Ashby, and show Ashby profiles so nobody is contacted twice. They can sit beside Ashby.

Buying nothing means reviewers read every application themselves. It also means no AI notice, opt-out handling or credits to manage.

Sources Every fact about Ashby and its alternatives comes from one of these addresses. Each line carries the date it was read.

https://www.ashbyhq.com/ai · Ashby AI · read 9 Oct 2026
https://docs.ashbyhq.com/ai-assisted-application-review · Ashby help, AI-assisted application review · read 9 Oct 2026
https://www.ashbyhq.com/product-updates/ai-assisted-application-review · Ashby product update, AI-assisted application review · read 9 Oct 2026
https://www.ashbyhq.com/product-updates/ai-model-upgrade-ai-assisted-application-review · Ashby product update, AI model upgrade · read 9 Oct 2026
https://www.ashbyhq.com/product-updates/candidate-opt-out-reporting · Ashby product update, candidate opt-out reporting · read 9 Oct 2026
https://docs.ashbyhq.com/ai-talent-rediscovery · Ashby help, AI Talent Rediscovery · read 9 Oct 2026
https://www.ashbyhq.com/product-updates/ai-talent-rediscovery · Ashby product update, AI Talent Rediscovery · read 9 Oct 2026
https://www.ashbyhq.com/product-updates/all-candidates-rediscovery · Ashby product update, All Candidates in Rediscovery · read 9 Oct 2026
https://docs.ashbyhq.com/application-review · Ashby help, application review · read 9 Oct 2026
https://docs.ashbyhq.com/auto-reject-applications · Ashby help, auto-reject · read 9 Oct 2026
https://docs.ashbyhq.com/ai-features-in-ashby · Ashby help, AI features · read 9 Oct 2026
https://www.ashbyhq.com/pricing · Ashby pricing · read 9 Oct 2026
https://docs.ashbyhq.com/ai-credits · Ashby help, AI credits · read 9 Oct 2026
https://docs.ashbyhq.com/best-practices-for-managing-ai-credits · Ashby help, managing AI credits · read 9 Oct 2026
https://docs.ashbyhq.com/seat-based-pricing · Ashby help, seat-based pricing · read 9 Oct 2026
https://docs.ashbyhq.com/billing-faq · Ashby billing questions · read 9 Oct 2026
https://www.ashbyhq.com/resources/terms · Ashby customer terms · read 9 Oct 2026
https://www.ashbyhq.com/resources/sla · Ashby service level agreement · read 9 Oct 2026
https://www.ashbyhq.com/resources/terms-ai-features · Ashby terms for AI features · read 9 Oct 2026
https://docs.ashbyhq.com/candidate-fraud-detection-overview-and-admin-settings · Ashby help, candidate fraud detection · read 9 Oct 2026
https://www.ashbyhq.com/resources/gdpr · Ashby GDPR · read 9 Oct 2026
https://www.ashbyhq.com/resources/security · Ashby security · read 9 Oct 2026
https://docs.ashbyhq.com/data-privacy-faq · Ashby data privacy questions · read 9 Oct 2026
https://docs.ashbyhq.com/how-do-i-export-candidate-data · Ashby help, exporting candidate data · read 9 Oct 2026
https://www.ashbyhq.com/downloadables/ashby-bias-audit-08-2024.pdf · FairNow bias audit of Ashby, August 2024 · read 9 Oct 2026
https://www.ashbyhq.com/customers/January-chose-Ashby-to-scale-from-series-b-and-beyond · Ashby customer story, January · read 9 Oct 2026
https://www.ashbyhq.com/customers/dave · Ashby customer story, Dave · read 9 Oct 2026
https://www.ashbyhq.com/customers/netgear · Ashby customer story, NETGEAR · read 9 Oct 2026
https://www.ashbyhq.com/blog/company/series-d · Ashby Series D post · read 9 Oct 2026
https://www.ashbyhq.com/blog/culture/ashby-one-2026-keynote · Ashby One 2026 keynote post · read 9 Oct 2026
https://docs.ashbyhq.com/juicebox · Ashby help, Juicebox · read 9 Oct 2026
https://docs.ashbyhq.com/sourcing · Ashby help, sourcing integrations · read 9 Oct 2026
https://support.greenhouse.io/hc/en-us/articles/41131886674075-Talent-Matching-FAQ · Greenhouse help, Talent Matching questions · read 9 Oct 2026
https://support.greenhouse.io/hc/en-us/articles/44504950876315-Talent-Matching-Data-Processing-FAQ · Greenhouse help, Talent Matching data processing · read 9 Oct 2026
https://support.greenhouse.io/hc/en-us/articles/51004677097627-Past-candidates-in-Talent-Matching · Greenhouse help, Past candidates · read 9 Oct 2026
https://www.workable.com/workable-ai · Workable AI · read 9 Oct 2026
https://www.workable.com/features · Workable features · read 9 Oct 2026
https://www.workable.com/pricing · Workable pricing · read 9 Oct 2026
https://help.workable.com/hc/en-us/articles/38381544828695-Using-the-Workable-Agent · Workable help, Using the Workable Agent · read 9 Oct 2026
https://investor.wedbush.com/wedbush/article/gnwcq-2026-3-13-workable-launches-agentic-ai-for-hiring-with-new-workable-agent · Workable Agent launch announcement · read 9 Oct 2026
https://www.gem.com/product/ai-app-review · Gem AI Application Review · read 8 Oct 2026
https://www.gem.com/product/ats · Gem ATS · read 8 Oct 2026
https://help.gem.com/external/gem-ai-rediscovery-overview-getting-started · Gem help, AI rediscovery · read 8 Oct 2026
https://help.gem.com/external/gem-ai-legal-privacy-compliance-faqs · Gem AI legal and privacy questions · read 8 Oct 2026
https://help.gem.com/databases/gem-help-center/gem-ats-knockout-questions · Gem help, knockout questions · read 9 Oct 2026
https://www.seekout.com/solutions/recruit/ · SeekOut Recruit · read 8 Oct 2026
https://juicebox.ai/peoplegpt · Juicebox PeopleGPT · read 8 Oct 2026
https://docs.juicebox.ai/talent-rediscovery · Juicebox help, Talent Rediscovery · read 8 Oct 2026