Workable AI Resume Screening: What It Actually Does

Short answer: Yes, and more directly than most of its rivals. Checked on August 17, 2026, Workable ships two separate things: a Screening Assistant that reads inbound resumes and is included in its Recruiting plans at no extra charge, and a paid Workable Agent add-on that evaluates candidates, sources passive ones and messages them, billed in AI credits at one credit per candidate screening. Workable is one of the few ATS vendors that both reads your inbound pile and publishes what it costs to do so.

Last updated August 2026.

Does Workable use AI to screen resumes?

Yes. Workable's Screening Assistant scans the resume attached to each applicant and produces a summary of how that person's skills and experience line up against the requirements of the job. It uses semantic matching rather than keyword matching, which means it is built to recognize that "managed a P&L" and "owned budget responsibility" describe the same thing even though they share no words. It draws on the profile summary, education, experience, skills and the attached resume file.

This puts Workable in a different position from several better-known names. Greenhouse deliberately refuses to score or rank applicants for role fit and argues publicly against black-box composite scoring. HireVue does not read inbound resumes at all, because every one of its AI products acts on something the candidate completes. Workable just reads the pile and tells you what it found.

The useful question is therefore not whether Workable does AI screening. It is which of the two products you are actually buying, what the metered one costs at your volume, and what the autonomous one is permitted to do without asking you.

Is the Workable AI Screening Assistant free?

The Screening Assistant is included in Workable's Recruiting plans, so basic AI resume screening does not cost extra on top of the subscription. That is genuinely unusual and worth crediting. Several vendors in this category treat resume scoring as a premium upsell or bury it in a quote-only enterprise edition.

What is not free is the Workable Agent. That is a separate add-on available on all three plans, and it is where the consumption pricing lives. Every paid account starts with 3,000 free AI credits to try it, after which you buy credit packs. So the honest answer to "is it free" is: the assistant is, the agent is not, and the difference between them is mostly about autonomy.

What the Workable Agent adds, and what it costs per candidate

Workable describes the agent as covering sourcing, screening and candidate communication. In its own words, the agent "searches the talent market, evaluates profiles against your job needs, sends personalized outreach, follows up to gather missing details, and advances candidates through your pipeline."

The pricing is metered per action, which makes it unusually easy to model. These are Workable's published credit costs and pack prices as of August 17, 2026, with the per-candidate cost worked out from them.

Action	Credit cost	Cost per action at $0.10/credit
Evaluation (candidate screening)	1 credit per candidate	about $0.10
Sourcing (passive candidate identified)	2 credits per candidate	about $0.20
Chat (candidate interaction)	10 credits per interaction	about $1.00

The credit packs are 5,000 credits for $600, 10,000 for $1,000, and 50,000 for $4,750. That works out to $0.12, $0.10 and about $0.095 per credit respectively, so buying in bulk saves roughly 20 percent. Since one evaluation is one credit, screening a candidate with the agent costs somewhere between about nine and a half cents and twelve cents depending on how much you buy at once.

Notice that a chat interaction costs ten times an evaluation. If you are modeling a conversational screening workflow where the agent asks each applicant a few follow-up questions, that is not a rounding error. Ten thousand applicants who each have one chat exchange is 100,000 credits, which at the best published rate is around $9,500.

How much does Workable cost per month, all in?

Workable publishes its subscription rates, which deserves credit. Checked August 17, 2026: Standard is $299 a month or $3,588 a year, Premier is $599 a month or $7,188 a year, and Enterprise is $719 a month or $8,628 a year. Annual billing saves about 20 percent. Active jobs are unlimited subject to fair usage limits, plans scale with employee headcount from 1 to 20 up through 1,000 and above, and there is a 15-day trial of the complete Standard feature set with no credit card required.

One packaging detail catches people out. Premier includes texting, video interviews, assessments, and performance and engagement by default. On Standard those are charged separately at roughly $89 to $109 a month each. Add three of them to Standard and you are past the Premier rate, so do that arithmetic before choosing the cheaper tier.

Putting subscription and credits together gives you a realistic monthly figure for agent-based screening at different volumes. Everything here is calculated from Workable's published numbers at the $0.10 credit rate.

Monthly applicant volume	Credits for evaluation	Credit cost	Plus Standard subscription
500 applicants	500	about $50	about $349/mo
1,000 applicants	1,000	about $100	about $399/mo
5,000 applicants	5,000	about $500	about $799/mo
25,000 applicants	25,000	about $2,375 at the bulk rate	about $2,674/mo

Two caveats so this table is not misread. First, the Workable subscription buys an entire applicant tracking system, not just screening, so these are not like-for-like rates against a screening-only tool. Second, if the included Screening Assistant is enough for your needs, the credit column is zero and you are just paying the subscription. The credits only apply to the agent.

Still, the shape of the curve is the thing to notice. Metered per-candidate screening is cheap at low volume and scales linearly with your applicant flow, which is exactly the opposite of what you want when a single hourly requisition draws 900 applications. Teams doing genuinely high-volume bulk resume screening should model their worst month, not their average one.

Does Workable AI auto-reject candidates?

Workable says a human stays in control. Its agent page states that "you stay in the loop while the agent handles time-intensive tasks, so you can spend more time on human decisions," and describes keeping you "in control at every step." That is the right posture and it matches what responsible vendors in this category now say.

The line worth reading twice is the one where the agent "advances candidates through your pipeline." Advancing is an action taken on a real person's application, and any system that can advance can usually also be configured to not advance, which is a rejection by omission. That is not a criticism of Workable specifically, it is the central governance question for every agentic hiring tool shipped in the last two years. Before you switch it on, be deliberate about exactly which actions an autonomous agent is allowed to take on its own and which require a person to click.

Get the answer in writing during the sales process, and get it at the level of specific configuration rather than reassurance. Ask what the agent can do without human confirmation on default settings, whether disposition can be automated, and what the audit trail records when the agent acts.

Does Workable rank candidates?

Workable's own materials describe evaluating profiles against your job needs and summarizing how each candidate's skills and experience match the requirements, with a ranking function to compare and sort candidates based on those evaluations. So there is ordering, not just a per-candidate summary.

What matters more than whether a number exists is what the number is computed against. Workable evaluates against the job requirements, which in practice means the job description you posted. Job descriptions are written to attract applicants, not to encode selection rules. They mix hard gates and nice-to-haves in the same paragraph, they list five years of experience when three would do, and they rarely state the one credential that actually disqualifies someone. An evaluation that inherits that document inherits all of it.

That is the gap explicit knockout criteria fill. Writing "must hold an active CDL Class A" as a hard gate, separately from "prefers night shifts" as a preference, produces a different and more defensible ranking than semantic similarity to a posting. It is also the difference a compliance reviewer will ask about later.

How Workable compares to the other ATS vendors on screening

The spread across major vendors is wider than most buying guides admit, and it is the most useful thing to understand before shortlisting. Each row here reflects that vendor's own published material on the date checked.

Vendor	Reads inbound resumes?	How screening is charged
Workable	Yes. Screening Assistant included; agent adds autonomy	Assistant included in plan. Agent metered at 1 credit per candidate
Greenhouse	Parses and filters, but deliberately does not score or rank for role fit	Not applicable, since it declines to score
iCIMS	Yes. Coalesce AI ranks on skills and experience	Included, though several agents and Copilot are optional
Workday	Yes, via HiredScore AI for Recruiting	Licensed separately from core Workday Recruiting
HireVue	No. Its AI acts on assessments and interviews candidates complete	Not applicable, and pricing is quote-only
Spark Hire	Yes. AI Resume Review inside its ATS product	Annual allowance: 6,000 reviews a year on Recruit Pro
SmartRecruiters	Yes. Winston Match ranks applicants and Winston Screen evaluates answers	Included in the edition that carries it, and pricing is quote-only
BambooHR	No native AI screening in its ATS at all	Not applicable. Per employee per month for the wider HR suite

Four genuinely different business models sit in that last column. Workable meters per candidate, Workday sells the AI as a separate license, Spark Hire caps reviews per year, and iCIMS includes ranking but makes agents optional. If you are cross-shopping, the feature checkboxes will look nearly identical and the invoices will not. The same question produces a yes at SmartRecruiters, where Winston Match ranks applicants, and a flat no at BambooHR, which has no native AI screening at all. Our Greenhouse vs Lever vs Workable comparison goes further into how the three suites differ outside of AI.

Where Workable AI screening leaves a gap

Three limits are worth planning around, and none of them are secret.

It only works on candidates inside Workable. If your system of record is Greenhouse, Lever, Bullhorn or an internal tool, Workable's screening is not available to you without migrating your ATS. That is a large project to undertake for one capability, which is why teams evaluating a Workable alternative are often not shopping for an ATS at all.

The cost scales with your problem. Per-candidate metering is elegant at 300 applicants a month and uncomfortable at 20,000, because the months when you most need screening are precisely the months it costs the most. Flat applicant allowances behave the other way around.

And the evaluation is anchored to your job description unless you do the work to separate hard requirements from preferences. That is not a software limitation so much as a process one, but it determines whether the ranking you get is defensible.

What to do with this

If you already run Workable and have not turned the Screening Assistant on, do that first. It is included, it costs nothing extra, and for a team with moderate application volume it may be the entire answer. There is no sense buying a second tool before using the one you have paid for.

If application volume is your actual problem, model the credit cost at your worst month before enabling the agent, and compare it against flat-rate screening layers that sit alongside your existing ATS. Prescreener publishes its rates for exactly this comparison: $199 a month for 1,000 applicants, $499 for 5,000, and $1,290 for 25,000, with no free plan. It reads every inbound application, applies the knockout and eligibility criteria you write for that specific requisition, asks your screening questions, scores role fit, and returns the pile ranked with a written reason under each candidate.

It also never rejects anyone. It ranks and flags, a recruiter opens a list that is already ordered, and a person makes every call. Candidates are told AI is used, the same criteria are applied to everyone, and results are bias-audited to support EEOC standards and NYC Local Law 144. What it does not do is verify licenses, run background checks, handle credentialing, check motor vehicle records, manage I-9 or E-Verify, or touch payroll. Those stay in the regulated systems you already run.

If you are pricing alternatives to a metered agent, our AI candidate screening software roundup sets out which vendors publish flat rates and which bill by the candidate.