Over the last month, we've reviewed AI interview platforms one at a time — HireVue on its own, Paradox against HireVue, five platforms against the same candidate, a startup-budget roundup. That approach is good for depth. It's bad for the question most people actually arrive with: given my hiring volume, budget, and risk tolerance, which category of tool should I even be looking at?
This piece answers that question directly. It's a full map of the AI interview market as it stands in July 2026 — every platform we've tested or researched this year, organised by what they actually are rather than how they market themselves, with pricing, fit, and the honest tradeoffs of each category.
Two things have changed materially since we started this series that make an updated map worth publishing now. First, Paradox was acquired by Workday, completed October 1, 2025, and Olivia is being folded into Workday's talent acquisition suite — changing its buyer profile from "standalone conversational AI vendor" to "Workday customer expansion." Second, a new class action against Eightfold AI, filed in January 2026 by former EEOC chair Jenny Yang and the nonprofit Towards Justice, has shifted the legal conversation from bias claims toward whether AI screening vendors are unlicensed consumer reporting agencies under the FCRA — a materially different and, in some ways, more dangerous theory for the industry. We cover what that means for each category below.
The Five Real Categories
Vendors market themselves as unique, but the market sorts cleanly into five functional categories. Almost every confused buying decision we've seen comes from comparing tools across categories that aren't actually competing with each other.
- Enterprise conversational & video suites — full-scale platforms built for six-figure hiring volumes, sold as part of a broader talent acquisition stack.
- Startup & mid-market async video — simpler, cheaper video screening tools built for teams making dozens, not thousands, of hires.
- Live conversational AI interviewers — AI conducts a real, adaptive spoken conversation rather than scoring pre-recorded answers.
- Technical screening specialists — code-first assessment platforms, increasingly scoring how candidates use AI tools, not just output.
- Interview intelligence layers — tools that sit alongside human interviews, recording and analysing rather than replacing the interviewer.
We'll go through each, with the specific platforms worth knowing.
Category 1: Enterprise Conversational & Video Suites
HireVue remains the default for large-volume hiring — the company has processed more than 70 million interviews across 700+ enterprise customers. As we found in our full HireVue review, pricing runs roughly $25 per interview with a $35,000/year minimum on the entry tier, scaling to $100K+ for enterprise contracts. It no longer scores facial expression — the NLP layer evaluates language content, competency-keyword matching, and response structure exclusively (a shift we detail in our guide to what AI interviews actually score).
Paradox (Olivia), now a Workday product, is a conversational recruiting assistant more than a pure interview tool — its core strength is candidate engagement and interview scheduling via chat, with structured screening layered on top. Pricing is not published; third-party estimates put mid-market deployments at $25,000–$60,000/year and enterprise contracts at $75,000–$150,000+, with implementation fees of $15,000–$35,000 on top. Our HireVue vs Paradox comparison still holds on the product mechanics, but the Workday acquisition changes the buying calculus: Olivia now makes most sense if you're already a Workday customer, and less sense as a standalone best-of-breed pick.
Eightfold AI is the outlier in this category — its core product is talent intelligence and candidate matching rather than interviewing per se, but it's included here because of how much it shapes who gets to an interview at all. As we covered in our full Eightfold review, the platform is currently facing a proposed class action alleging it scraped data on over a billion workers and scored applicants without disclosure — notably, the suit doesn't allege the algorithm is biased, it alleges the operation was secret, framing Eightfold as an unlicensed consumer reporting agency under the FCRA. Combined with the earlier Mobley v. Workday ruling that a vendor can be legally treated as an employer's "agent," this is the most important legal development in the category this year, and it applies pressure to every vendor in this tier, not just Eightfold.
Organisations hiring in the thousands per year with dedicated TA operations teams and existing HRIS/ATS infrastructure to integrate against. If you're evaluating this tier, budget for legal review of the vendor's data handling and disclosure practices alongside the product demo — not after signing.
Category 2: Startup & Mid-Market Async Video
As detailed in our startup AI interview tools roundup, this tier exists because enterprise suites are overkill — in price and implementation complexity — for teams making ten to fifty hires a year. The shortlist:
- Spark Hire — $249/month, deepest ATS integrations, best for teams already doing structured hiring at some scale.
- Willo — $249–$399/month, strongest candidate experience and GDPR tooling, best for EU/UK hiring.
- myInterview — from $49/month, lowest price floor, AI-generated candidate summaries for async collaboration.
- Manatal — $15/user/month, an ATS with AI scoring built in rather than a standalone interview tool, best for teams that don't have a proper pipeline system yet.
Nothing material has shifted in this tier since our original roundup — it remains the most stable, lowest-risk part of the market, largely because none of these vendors are making aggressive claims about bias detection or predictive scoring. They screen; they don't purport to judge character or culture fit, which keeps them out of the legal crosshairs affecting the enterprise tier.
Category 3: Live Conversational AI Interviewers
This is the newest and fastest-moving category — platforms where the AI conducts a genuine adaptive conversation rather than scoring pre-recorded answers to fixed prompts.
InterviewFlowAI charges roughly $0.99 per interview with no monthly minimum, making per-hire cost highly predictable for low-volume hiring — a structurally different model from the subscription tiers above it. Ray by Diyam AI, which we tested directly against the field in our Ray vs the field comparison, differentiates on adaptive technical depth — probing follow-up questions based on the specifics of a candidate's claimed experience rather than a fixed script. Braintrust AIR has entered this category more recently, running real-time adaptive video interviews that evaluate technical skill, communication, and behavioural signal simultaneously.
The honest limitation across this entire category, which we documented at length in why AI interviewers should feel like real ones — and why most don't and again in our five-platform head-to-head test: adaptive follow-up logic in 2026 is still substantially rule-based rather than genuinely reactive. When a candidate makes a claim that should trigger scepticism — an inflated ownership claim about a project, for instance — most of these systems ask a generic clarifying question about the tech stack rather than directly challenging the claim the way an experienced human interviewer would. The category is closing that gap quickly, but it isn't closed yet.
Category 4: Technical Screening Specialists
Codility and CodeSignal remain the standard for pure coding assessment — correctness, efficiency, and test pass rate under time pressure. Codility has also become the accessibility leader in the category, with WCAG 2.2 compliance ahead of most competitors.
The category's most significant 2026 development is HackerEarth OnScreen, which launched publicly on April 14, 2026. It conducts a full AI-led first-round technical screening — a role-calibrated conversation that adapts to candidate responses and produces a structured scorecard, effectively merging the "technical assessment" and "autonomous AI interviewer" categories into one product. It's worth watching closely: if it performs as advertised, it puts pressure on both Category 3 and Category 4 vendors simultaneously.
The broader trend across this whole tier, which we flagged in our guide to what AI interviews actually score: assessment platforms are increasingly scoring how candidates use AI coding assistance during the task, not just whether the final code works. That's a genuine shift in what "technical screening" measures, and it's still early — most candidates are not yet aware their AI-tool usage pattern is part of the signal being captured.
Category 5: Interview Intelligence Layers
This category doesn't replace the human interviewer — it sits alongside one. BrightHire and Metaview lead here, recording, transcribing, and analysing human-led interviews to generate structured notes, flag inconsistencies across a candidate's answers over multiple rounds, and reduce the note-taking burden on interviewers. Radancy has positioned itself specifically around soft-skills and culture-fit evaluation within this layer.
This is arguably the lowest-controversy category in the market — it doesn't make an autonomous pass/fail decision, which keeps it outside most of the current legal scrutiny aimed at scoring and screening vendors. For organisations wary of the Eightfold-style FCRA exposure but still wanting AI leverage in hiring, this category is worth serious consideration as a lower-risk entry point.
Full Comparison Table
| Platform | Category | Starting Price | Best Fit |
|---|---|---|---|
| HireVue | Enterprise suite | $35K/yr min. | High-volume enterprise hiring |
| Paradox (Olivia) | Enterprise suite | ~$25K–$60K/yr | Existing Workday customers |
| Eightfold AI | Enterprise suite / matching | Enterprise, undisclosed | Large-scale talent matching (legal risk — review carefully) |
| Spark Hire | Startup / mid-market | $249/mo | Structured hiring, strong ATS needs |
| Willo | Startup / mid-market | $249/mo | Candidate experience, EU/UK hiring |
| myInterview | Startup / mid-market | $49/mo | Early-stage, budget-conscious |
| Manatal | Startup / mid-market (ATS) | $15/user/mo | Needs full pipeline, not just interviews |
| InterviewFlowAI | Live conversational AI | ~$0.99/interview | Low-volume, per-hire pricing |
| Ray by Diyam AI | Live conversational AI | Contact for pricing | Technical roles, adaptive depth |
| Braintrust AIR | Live conversational AI | Contact for pricing | Enterprise, multi-signal live video |
| Codility / CodeSignal | Technical screening | Varies | Implementation-heavy roles |
| HackerEarth OnScreen | Technical + autonomous AI | Contact for pricing | First-round technical screening at scale |
| BrightHire / Metaview | Interview intelligence | Contact for pricing | Augmenting human interviews, lower legal risk |
What This Means for the "State of AI Hiring" Picture
Our State of AI Hiring 2026 report and our research summary on AI hiring bias both flagged the same underlying tension: adoption is accelerating faster than transparency practices are maturing. This comparison reinforces that finding at the product level. The categories with the least legal exposure right now — startup async video and interview intelligence layers — are also the categories making the narrowest, most honest claims about what they measure. The categories under the most legal scrutiny — enterprise matching and scoring platforms — are the ones making the broadest claims about predictive power.
That correlation is not a coincidence, and it's a useful heuristic for buyers: the vendors most confident about what their AI can predict about a candidate tend to be the ones facing the most questions about how they arrived at that confidence.
How to Actually Choose
Stripped of vendor positioning, the decision tree is simpler than the market suggests:
- Hiring 1,000+ people a year with existing HRIS infrastructure: Category 1, but budget time and legal review for vendor data practices given the current litigation climate.
- Hiring 10–100 people a year, budget-conscious: Category 2. This remains the lowest-risk, most stable tier in the market.
- Hiring for technical roles where depth of experience matters more than volume: Category 3, with the caveat that adaptive follow-up is still maturing — don't expect it to catch every inflated claim yet.
- Screening for implementation ability specifically: Category 4, and ask directly how (or whether) AI-tool-usage scoring factors into the result.
- Want AI leverage without an autonomous scoring decision: Category 5 — the lowest-controversy entry point into AI-assisted hiring today.
The right AI interview platform isn't the one with the most features or the biggest name. It's the one whose category actually matches your hiring volume, your risk tolerance, and what you're honestly trying to measure.
This is the twelfth piece in our AI recruitment research series — covering enterprise tool reviews, startup roundups, live testing, candidate-side research, and now a full market map. For methodology on any individual platform above, the reviews linked throughout this piece go deeper than we could here. Sharingan AI evaluates recruitment technology independently, without vendor sponsorships or affiliate relationships.