Workable vs Greenhouse

Workable tells you what it costs before you talk to anyone, bundles an HRIS into the same subscription, and as of 20 July 2026 exposes 94 MCP tools to any AI assistant on every plan at no extra charge. Greenhouse publishes no number at any size, does recruiting only, and gates its own MCP behind an open beta on its Core, Plus, and Pro tiers — but it is the system of record the rest of the recruiting ecosystem builds against, and its 2026 AI wave was explicitly designed to feed structured hiring rather than route around it. The usual framing of this pair — cheap SMB tool versus expensive enterprise ATS — is wrong above roughly 100 employees, and the pricing section below shows where it stops being true.

Where Workable wins

• It is one subscription for recruiting and HR. The same plan carries an HRIS, onboarding, time-off management, time tracking, e-signatures, and payroll preparation alongside the ATS. For a 60-person company that would otherwise buy Greenhouse plus BambooHR, that is two contracts collapsed into one, and employee records that don’t need an integration to stay in sync. Greenhouse does not sell an HRIS and does not intend to.
• The MCP surface is the deepest any ATS has shipped. Workable launched its MCP server on 13 May 2026 with 57 tools and expanded it to 94 on 20 July, covering jobs, candidates, pipeline stages, offers, requisitions, employees, time off, and the full performance-review lifecycle. It handles read and write, connects over OAuth2 to Claude, ChatGPT, or any MCP client, and access is scoped to the signed-in user’s own permissions and assigned jobs. Most of it is on every plan at no added cost; only Advanced Candidate Search and Advanced Employee Search require Premier or Enterprise.
• The agent runs the top of the funnel, not just a step of it. Workable Agent, launched 13 March 2026, takes an intake conversation, builds an ideal-candidate profile, sources against 400M+ profiles, sends personalized outreach, chats with candidates to fill gaps, scores against the spec, and hands over a shortlist. It re-scores as new information arrives, logs every action, and Workable states it excludes protected characteristics from scoring by design. See autonomous AI recruiter for what that category does and does not cover.
• You can price it in a browser tab. Three tiers, banded by employee count, with the number on the page. Greenhouse’s tier page names Core, Plus, and Pro and then says pricing is customized — which means a demo call before you know whether it is even in range.
• Distribution is included, not integrated. 200+ job boards, careers page builder, Indeed Easy Apply, LinkedIn Recruiter System Connect, and resurfacing of past candidates ship in every plan.

Where Greenhouse wins

• Structured hiring is the product, not a feature of it. Interview kits, scorecards, and source-of-hire reporting are first-class objects, and the AI wave feeds them: the Greenhouse Notetaker (mid-July 2026) maps AI interview notes back onto the specific scorecard questions the kit defines. Workable ships interview kits and scorecards too — the difference is enforcement and reporting depth, not existence. If your hiring quality problem is interviewer discipline at scale, read structured interviewing and price that gap honestly.
• MCP access is governed at the org level, not the user level. Greenhouse MCP, announced 7 May 2026 with StubHub and Komodo Health as design partners and in open beta since, requires an organization to approve scopes in Dev Center → MCP Access before any user connects. Every call then inherits that user’s existing Greenhouse permissions and lands in an audit trail. Workable’s model — permissions follow the individual, no org-level scope gate — is faster to adopt and gives a security review less to hold onto.
• The integration marketplace is the moat. 450+ certified integrations means whatever assessment, background-check, scheduling, or sourcing tool you pick, it was built against Greenhouse first. Workable’s all-in-one thesis is that you need fewer of them, which holds until the one you need is the one Workable doesn’t build.
• Fraud and identity controls sized for the application flood. Greenhouse’s own figure is that applications are up 412% since 2023 while open roles stayed roughly flat. Real Talent adds talent matching in Core and fraud and spam detection in Plus; Greenhouse also acquired voice-AI interviewer Ezra AI Labs in May 2026, putting screening-stage interviewing inside the same vendor as the scorecard.
• It survives becoming a public company. Audit log, enterprise data configuration, developer sandbox and sync, and the permissioning and approval flows that a controls review asks for are Pro-tier features that exist. Workable’s ceiling is real, and it is somewhere in the low thousands of employees.

Pricing reality

Workable publishes per-headcount pricing that scales with total employees, not recruiters or hires. At the 1–20 employee band, Standard is $299/month ($3,588/year), Premier $599/month ($7,188/year), Enterprise $719/month ($8,628/year). At the 101–200 band the same tiers are $949/month ($11,388/year), $1,759/month ($21,108/year), and $2,109/month ($25,308/year). Add-ons are separate and not small: at that band, Texting is +$269/month, Video Interviews +$409/month, Assessments +$209/month.

Greenhouse publishes nothing. Against Vendr’s verified-purchase data as of today, its median annual contract is $26,646 across 866 purchases, ranging $10,215 to $75,125, with buyers averaging 16% off list.

That comparison inverts as you grow. At 20 employees, Workable Standard at $3,588 is roughly a third of the cheapest Greenhouse deal Vendr has recorded. At 200 employees, Workable Premier at $21,108 is about 21% under the Greenhouse median — a rounding error against the cost of migrating an ATS. The price argument for Workable is decisive under 50 employees and mostly spent by 200. Above that, you are choosing on process and ecosystem, which is the choice Greenhouse wants you to make.

The AI line item works differently on each side. Greenhouse folds the Notetaker into Core and routes MCP access and pricing through your account team. Workable meters its agent in AI credits: 1 credit each time a candidate is evaluated against the profile, 2 credits each time a passive candidate is sourced into the pipeline, 10 credits each time a candidate interacts with the chat. Every paid account gets 3,000 free credits; bundles are 5,000 for $600 ($0.12/credit), 10,000 for $1,000 ($0.10), and 50,000 for $4,750 ($0.095). Work an example: a quarter with 2,000 applicants evaluated, 300 passive candidates sourced, and 200 candidate chats costs 2,000 + 600 + 2,000 = 4,600 credits, or about $460 at the 50,000-credit rate and $552 at the 5,000-credit rate. The chat action is 10x the evaluate action, so conversational screening — not volume screening — is what moves that bill.

Watch-outs, and what to do about each

• Workable AI credits expire one year from purchase and are non-refundable. Guard: buy the 5,000-credit bundle first, run one full req cycle, and derive your annual size from measured consumption before committing to the 50,000 tier.
• Chat is the credit sink. A high-volume inbound role that routes every applicant through agent chat burns 10 credits a head. Guard: configure chat to trigger only after the evaluate step passes, and cap it per job.
• Workable’s price is indexed to headcount, not hiring. A company that doubles its employees while hiring flat still pays the next band. Guard: model the quote at your 18-month headcount plan, not today’s, and negotiate the band transition in the contract.
• The Workable free trial excludes the Agent. The 15-day trial covers the Standard feature set only, so the part you are actually evaluating isn’t in it. Guard: make a scoped agent pilot on one live req a written condition of signing.
• Greenhouse MCP is open beta and legacy tiers are excluded outright. A long-tenured customer on an older plan may need a tier move to get agent access at all. Guard: confirm in writing that your subscription is Core, Plus, or Pro, and get MCP access and its price named in the order form rather than described on a call.
• “Custom pricing” means your renewal is a negotiation you have to staff. Vendr’s data shows buyers averaging 16% off list, which is money that only exists if someone runs the process. Guard: request Core, Plus, and Pro quotes simultaneously and benchmark against the $10,215–$75,125 observed range before the first counteroffer.

Verdict

• Pick Workable when you are under roughly 200 employees and have no HRIS you’re committed to; when the recruiting workload is bursty and there is no dedicated TA ops owner to run a process-heavy system; when you want agent access to hiring and HR data on every plan without an org-level enablement project; or when a published price is what makes the purchase approvable at all.
• Pick Greenhouse when interviewer discipline is the thing you’re protecting and scorecard compliance has to be reportable; when you hire enough across panels that source-of-hire and stage-conversion reporting drive real decisions; when your stack depends on integrations only the largest marketplace carries; or when a security review will insist that AI access to candidate data be org-approved and audited rather than user-scoped.
• Pick neither when the constraint is elsewhere. If you want an all-in-one with deeper native analytics than either, Ashby is the more direct competitor to both — see Ashby vs Greenhouse. If your HRIS gravity is already Workday, the suite argument beats both. If you want Greenhouse’s shape at a lower price point, Lever is the usual next call (Greenhouse vs Lever). The full field is in best ATS platforms 2026 and alternatives to Greenhouse.

Default pick: Workable. A team that genuinely cannot decide between these two is almost always below the scale at which Greenhouse’s process rigor pays for itself, and Workable is the only one of the pair that will quote you without a call, covers HR as well as recruiting, and ships agent access on the entry tier. Switch to Greenhouse when a specific condition forces it: a dedicated TA ops owner exists to run the process, panel interviewing across 100+ hires a year makes scorecard reporting load-bearing, or an integration you depend on only certifies against Greenhouse. What should not decide it is the headline price — by 200 employees the gap is about 21%, and at that point you are paying for a hiring process, not a license. If high inbound volume is the real problem, AI resume screening and the high-volume recruiting stack matter more than which of these two you sign.