Maki People

What it is

Maki People runs the top of the hiring funnel — screening, assessment, interviewing, scheduling — through five named agents that sit on top of an existing ATS rather than replacing it. Shiro screens for skills across 300+ competencies. Mochi runs conversational voice interviews in 45+ languages, in the browser or over the phone. Ken delivers role-calibrated cognitive, behavioral, and technical assessments across 400+ skills. Tomo co-pilots live human interviews with real-time structuring and transcript analysis. Kumi handles interview scheduling. The vendor’s own line — “the intelligence running your hiring, end to end” — is the reposition: Maki sells the funnel stage now, not the test.

In April 2026 Maki added an AI proficiency assessment inside Shiro and Mochi at no extra charge, scoring candidates on four dimensions — AI tool agility, human-tool interactivity, ethical use, and data and algorithmic literacy — delivered through situational judgment tests and structured behavioral grids. The score lands in the candidate profile alongside the Shiro and Mochi outputs.

Vendor-reported results: 45% faster hiring, 50% lower turnover, 250,000+ recruiter hours saved, 92.4% positive candidate feedback. Named customers include Deloitte, PwC, Capgemini, BNP Paribas, Generali, ASOS, Nespresso (49% turnover reduction), Grant Thornton, and Forvis Mazars (60% screening-time reduction). Maki raised a $28.6M Series A in January 2025 led by Blossom Capital with DST Global participating, on 300%+ revenue growth in 2024.

Why it shows up in recruiting stacks

• Adaptive conversation instead of a fixed question bank. Harver and TestGorilla ship fixed instruments — situational judgment tests, timed skill tests — scored identically for every candidate. Mochi and Shiro conduct a dialogue and follow the candidate’s answers. Where drop-off at the assessment step is the binding constraint — frontline retail, BPO, graduate intake — that format change is the reason to look at Maki at all.
• An overlay on 30+ ATS platforms, not a parallel system. Native connectors for iCIMS, Workday, and SAP SuccessFactors, plus 30+ more including Greenhouse, Lever, SmartRecruiters, Workable, Bullhorn, Teamtailor, Avature, and Eightfold. Agents are triggered from the ATS and scores sync back into it, so the recruiter’s decision surface does not move.
• 45+ languages in the voice agent. For a European enterprise hiring across a dozen markets, Mochi removes the need to staff bilingual screeners per market. Harver does not match this at conversational depth, and it is the clearest reason Maki’s logo wall skews to pan-European enterprises.
• NYC LL 144 audits by a named auditor, with dates. Holistic AI audits Mochi and Shiro annually; the most recent audits completed 23 March 2026, testing gender and race/ethnicity at both standalone and intersectional levels. For a US federal contractor or an EU AI Act-scoped buyer, a named auditor and a dated report is the difference between a vendor questionnaire that clears and one that stalls in legal review.

Pricing reality

Maki publishes no prices. The pricing page states a credit-based model only — “Pay only for what you use, candidate interactions, not licenses” — with credits consumed per candidate interaction and per agent activated, and the tier set by functionality and organization size at quote time. There is no free plan, no published floor, and no self-serve entry: every path ends at “Get a personalized quote.”

Third-party triangulation puts real contracts at $30,000-$100,000+ per year (HeroHunt vendor analysis, July 2026), inferred from category positioning rather than sourced from Maki. The ~$3,000/year figure still circulating from 2023 Capterra data predates the Series A and the five-agent product; treat it as dead, not as a floor. Budget five figures annually for any deployment large enough to justify the integration work.

Against what the alternatives publish: TestGorilla lists Free at $0 (10 credits/month), Core at $142/month billed annually ($1,704), and Plus from $400/month. Sapia.ai runs about $1 per interview on a roughly $100/month minimum. Harver is quote-gated, but third-party data puts entry near $5,000/year, mid-market at $8,000-$15,000, and enterprise above $20,000. Maki’s floor sits above all three, and that fact settles most evaluations before feature comparison starts.

Best for

Enterprise TA leaders running 500+ hires a year across multiple countries — retail, professional services, financial services, BPO — where the recruiter-to-req ratio makes manual first-round screening impossible and the candidate pool is multilingual. The use case where Maki compounds is early-careers cohort hiring: thousands of applications inside a two-week window, standardized structured assessment, automated scheduling, and multilingual candidate contact, all before a recruiter opens a file.

Do not buy Maki under ~100 hires a year — at that volume the credit model prices above TestGorilla’s $1,704/year Core plan for a capability you use in bursts. Skip it if your hiring concentrates in engineering roles graded on code: Ken covers cognitive, behavioral, and technical dimensions, but HackerRank and CodeSignal own code-challenge grading. Skip it if your TA ops team needs to build new role templates without vendor involvement — self-serve customization for non-standard roles is the most consistent complaint in G2 reviews.

Versus the alternatives

Harver is the installed-base incumbent in volume pre-employment assessment, with 450+ validated situational judgment tests built for frontline hiring and the deepest industrial-organizational psychology bench in the category. Pick Harver when the buying committee is led by IO psychology and validation evidence, and static SJTs cover your role types. Pick Maki when you need adaptive conversation and multilingual voice.

HireVue is the other incumbent by installed base, at roughly $35,000/year on third-party data, and the low-risk procurement answer at Fortune 500 scale. Pick HireVue when structured video interviewing plus assessment inside one long-standing vendor relationship matters more than agent orchestration. Pick Maki when the scope is the whole top of the funnel rather than the interview step.

TestGorilla is the transparent-pricing option and the place to start below enterprise scale: published tiers, a self-serve test builder, 400+ tests, and a conversational AI video interview capability shipped in October 2025 that competes with Mochi directly. Pick TestGorilla under 500 employees, or whenever a published price is a procurement requirement.

Sapia.ai is the fastest-growing conversational entrant — chat-based structured interviewing with published bias reporting, at a per-interview price you can model exactly. Pick Sapia when you want conversational screening without the assessment and scheduling agents attached.

If none of them fit — you need code grading and a published price at enterprise volume — the honest answer is a two-vendor split: HackerRank or CodeSignal for technical grading, TestGorilla for everything else, and no orchestration layer until hiring volume pays for one.

Watch-outs

• Pre-apply screening turns Mochi into a selection procedure. Maki’s Recruitics AdaptiveApply integration scores behavioral, situational, and linguistic performance before a candidate formally applies, and candidates who clear a threshold flow to the next stage. Scoring linguistic performance implicates national-origin (accent) and disability (speech difference, autism spectrum) exposure, and “a human makes the final call” does not cure it when the machine decides which humans get seen. The March 2026 Holistic AI audits cover gender and race/ethnicity; disability and veteran-status testing is not published. Guard: if you are a US federal contractor, do not deploy Mochi pre-apply — run it post-application only, get the applicant-definition guidance Maki committed to publishing in writing before you sign, and make the disability-status audit a dated contract deliverable.
• Credit consumption is not forecastable at procurement. Credits burn per candidate interaction and per agent activated, so a calendar with a graduate-intake spike costs a multiple of the average month, and no published unit rate exists to model against. Guard: negotiate a per-candidate cost ceiling alongside the annual credit commitment, get rollover terms for unused credits in writing, and cap how many agents a template change can activate without a new PO.
• No public API and no MCP server. Maki’s integrations are ATS connectors triggered from the ATS; the integrations page names no public API, no webhooks, and no LLM access path. Agent output is reachable where the connector writes it and nowhere else, so any warehouse-side analysis of screening decisions depends on the ATS record. Guard: before signing, confirm in writing that structured scores, transcripts, and audit logs export to your warehouse on a schedule rather than only to the ATS, and hold the final implementation payment against that delivery.
• Non-standard role templates route back to the vendor. G2 reviewers report that assessments for specialized roles require Maki’s team rather than self-serve configuration, which adds lag exactly when an unplanned req opens. Guard: during evaluation, build one non-standard role assessment yourself with no vendor help. If you cannot, negotiate a template-turnaround SLA with a committed response time into the contract before signing.