Nothing has been called PredictiveHire since 2022. The company rebranded to Sapia.ai that year, choosing the name as a contraction of Homo Sapien plus AI, ahead of its push into the UK and USA - PR Newswire. If you arrived here typing the old name, that is worth knowing first, because a lot of the PredictiveHire-era information still circulating (including the price) is either stale or was never true.

Sapia.ai is an Australian-built text-based AI interview platform for high-volume hiring. It sits between your job ad and your shortlist: candidates answer a handful of written questions in a chat window, and the platform ranks them. This page covers what it actually does, what it actually costs (the honest answer is more useful than the fake rate card you will find elsewhere), where it works, where it is drawing fire, and what to use instead.

1. The core technology

Sapia's flagship product is the Smart Interviewer, a structured chat interview. Sapia works with the client to define the traits a role needs, then builds "a structured interview with our AI Smart Interviewer, with the five questions needed to get the talent you're after" - Sapia.ai. Candidates get a link from the job ad and reply in writing. There is no camera and no timer, and they can leave and come back. Natural language processing then infers soft skills, behavioural traits and communication signals from the text, returns a ranked shortlist to the employer, and sends personalised feedback to every candidate, including rejected ones.

One correction worth making, because this site got it wrong before and the error is common: the Smart Interviewer is not called Phai. Phai is a separate Sapia product, an AI career coach aimed at job seekers and at internal mobility for existing employees, delivered via API into a chat UI. It is a different thing sold to a different buyer for a different reason. If you are evaluating Sapia for hiring, the Smart Interviewer is the product you are buying. The wider platform also includes Jas (an interview builder), Tia (a recruiter co-pilot), scheduling and an analytics dashboard.

The no-camera decision is the genuinely interesting one, and it is not just an ethics posture. Video interview tools carry accessibility and bias problems that text does not: no lighting, no bandwidth, no accent, no face. Text is also extremely cheap to run at a million interviews, which, as the next section shows, is the only reason Sapia's pricing model works at all. The product design and the commercial model are the same decision.

What it claims to do well:

1. Blind screening: the AI scores written responses without names, gender or educational background attached, which removes some of the obvious proxies for bias. Treat "eliminates bias" as marketing rather than fact: a language model scoring written English has its own well-documented sensitivities, and Sapia's own answer to this is auditing, not immunity.
2. Volume throughput: this is the real selling point. Automating first-round screening is what lets a retailer process a funnel it could never phone-screen.
3. Candidate feedback: every applicant gets a personality-based report, which is rare and is a genuine differentiator. It is also, as we will see, the part generating the loudest complaints.
4. Multilingual coverage: the interview runs in eight-plus languages including Spanish, French, German, Dutch, Danish, Finnish, Italian and Swedish.

Notice what is absent from that list. Sapia does not source candidates, and five written questions will not assess a senior engineer or a finance director. This is a screening layer for front-line, high-volume roles, and it is sold as exactly that.

2. The price

Sapia.ai charges per hire, not per interview, and publishes no rate. Its pricing page is unambiguous: "Sapia.ai's pricing structure works on a per-hire basis. You can interview and assess as many people as you want, without limits, and you'll only pay for the people you end up hiring" - Sapia.ai pricing. The rate is quoted against your annual hiring volume, and Sapia states the product is "designed to best serve companies that hire more than 500 people per year."

Be careful what you read elsewhere on this. There is a widely repeated set of Sapia "plans" (roughly $1 per interview, a $500/month tier and a $2,000/month enterprise tier) that appear on directory sites and get recited confidently by chatbots. Those plans do not exist. They are on no Sapia page and in no Sapia document, and they contradict the per-hire model in Sapia's own words. The "starting from $1" figure on aggregators like GetApp is a placeholder in a form field, not a price. There is no public Sapia rate card, and anyone quoting you one is guessing.

The per-hire model matters more than any number, because it decouples your screening cost from your applicant volume. Woolworths, Sapia's flagship account, interviews roughly 1 million people a year and hires about 50,000 - Sapia.ai case study. That is a 20:1 funnel. At even a dollar per interview it would cost a million dollars a year before a single hire. Priced per hire, the other 950,000 conversations are free. If you have a wide top of funnel (retail, hospitality, contact centres, seasonal peaks), that is the entire appeal.

Two questions to take into the call. First: is there a minimum annual commitment? A per-hire price with a floor under it is a subscription in a costume, and the floor is where your risk sits. Ask what a hiring freeze costs you. Second: what does a custom ATS integration cost? Standard integrations are included, but Sapia notes a custom one may carry a fixed fee. Our fuller breakdown of the model is in Sapia.ai recruiting pricing.

3. Adoption, and the backlash

Sapia's customer list is genuinely enterprise: Woolworths Group, Bunnings, Qantas, Kmart Group, Starbucks, David Jones, Holland & Barrett, Transavia and LNER. The company says over 10 million people have been through the interview experience. Its lead outcome case is Holland & Barrett, reporting store turnover falling from 74% to 15% within three months.

Read those numbers as directional rather than proven. A company that overhauls its screening usually changes several other things at once, and there is a disclosure worth knowing on the flagship reference: W23, Woolworths Group's venture arm, co-led Sapia's $17 million Series A with Macquarie Capital, announced 22 November 2022, at which point Woolworths had already been a customer for 12 months - PR Newswire. The headline case study describes an investor deploying its own portfolio company. That is not the same evidential weight as an arm's-length buyer, and you should ask for a reference customer that is not on the cap table.

The bigger update since this page was first written is that candidate sentiment has turned into a live PR problem. Australian jobseekers facing Sapia-powered interviews at Woolworths and Bunnings called the process "dehumanising," with one telling reporters "if they were actually interested in hiring you, they'd speak to you in person" - Yahoo Finance. Through early 2026 the complaints have centred on the feedback feature specifically, with applicants describing the AI's personality read-outs as inaccurate and worse: a 16-year-old was told his self-belief could alienate others - NZ Herald. The satisfaction scores in Sapia's sales deck and a public backlash can both be true at once, and the gap between them is usually the difference between the average candidate and the rejected one. If your employer brand is part of your hiring advantage, that trade-off is a real line item, not a footnote.

One more thing to price in on timing: AI used for recruitment and candidate selection is high-risk under the EU AI Act, with the compliance deadline now 2 December 2027 following the Digital Omnibus. The delay is not a repeal. If you hire in the EU, whether the vendor carries the conformity and human-oversight burden or hands it to you is a contract question worth raising while you still have signing leverage.

4. Alternatives

The right alternative depends on why Sapia does not fit, so match the option to the reason rather than to a feature grid.

If you want the same job done, with more automation: Paradox. The closest competitor for front-line volume hiring, and the one Sapia builds its own comparison page against. Paradox is a conversational hiring assistant that is stronger on scheduling and end-to-end automation, and Workday completed its acquisition of Paradox on 1 October 2025. That is an advantage if you are already a Workday shop and a lock-in risk if you are not. Sapia's counter-pitch is that it stays ATS-agnostic.

If you want assessment breadth: HireVue or Harver. Both are broader than Sapia (video, gamified and cognitive assessments) and both license conventionally rather than per hire. Pick them when you want a wider assessment battery. Pick Sapia when you specifically want text-only screening and an unlimited funnel.

If the problem is that nobody applied: a sourcing tool, not a screening tool. Sapia assumes a thousand applicants and picks the best of them. If your requisitions are starving, screening software is the wrong category entirely and no amount of it will help. HeroHunt.ai works the opposite end of the funnel, searching a large candidate pool and running first-touch outreach automatically.

If you hire fewer than 500 people a year

This is where most people reading a PredictiveHire page actually land, and Sapia more or less says so itself. Below its volume threshold the economics stop working: the per-hire rate negotiated for a 20,000-hire retailer is not on offer to you, and a dedicated AI interviewer is a lot of machinery for 80 hires a year. The realistic move is an applicant tracking system with AI shortlisting built in, which addresses the actual problem (too many applicants, no structure, no pipeline visibility) at a fraction of the cost, with a price you can read before you talk to a salesperson.

In that bracket Manatal is the one we would start with, on a narrow and specific merit: it is the closest thing to the opposite of Sapia's commercial model. Sapia negotiates every quote and will not name a number until it knows your hiring volume. Manatal publishes a rate card, runs a trial without a credit card, and scores and ranks inbound applicants automatically, which is the 80% of the job most sub-500 employers were trying to solve when they went looking for an AI interviewer. It does not conduct interviews, and that gap is the whole point of the comparison: if you genuinely need 4,000 structured conversations, you need Sapia, not a cheaper ATS.

The bottom line

PredictiveHire is Sapia.ai, it charges per hire with unlimited interviews, it publishes no rate, and it is built for employers hiring 500+ people a year into high-volume front-line roles. Within that niche the product is well matched to the problem, and the per-hire model is genuinely kinder to a wide funnel than anything billed per assessment. Outside it, the fit falls apart quickly.

Walk into the conversation with three numbers of your own: your annual hires, your applicant-to-hire ratio, and your current loaded cost per screen. Without the third one you cannot evaluate the quote and you will negotiate on vibes. Ask about the minimum annual commitment, ask for a reference that is not an investor, and weigh the candidate-experience reporting honestly against the satisfaction scores in the sales deck. And if you hire fewer than 500 people a year, the more useful question is not what Sapia would charge you, but whether you needed an AI interviewer in the first place.

Written by Yuma Heymans (@yumahey), who built HeroHunt.ai and spends most of his time on the economics of the hiring funnel: what screening really costs per candidate, and which parts of it are worth handing to a machine.

Sapia.ai does not publish pricing. Every figure on this page is sourced to Sapia's own pages, its investors' announcements, or published reporting. The per-hire rate is deliberately not estimated, because no primary source for it exists.