Vertical enterprise AI keeps pricing gated
100 of 338 corpus companies hide their pricing (~30%) — still a minority, but the gating clusters where buying is top-down. Vertical-knowledge AI is the densest cluster: 27 of the 61 vertical-SaaS companies gate (44%, well above the ~30% baseline), spanning legal (Harvey, Spellbook, Legora, CoCounsel), enterprise search (Glean, Hebbia), clinical (Nabla, Suki, Abridge, Ambience), and contact-center (Observe.AI, Uniphore, PolyAI). Billing infrastructure (11 gated) and outbound sales-tech (11x, Nooks, Apollo) round it out. The buying motion, not verticality alone, forces gating.
What's happening — and why
What's happening: most AI companies publish their prices — 238 of the 338 corpus companies (~70%) post a public number. The 100 that gate or sales-only their pricing concentrate in a handful of segments, and vertical-knowledge AI is the densest: 27 of the 61 vertical-SaaS companies (44%) show no number at all and route you to a sales team — legal (Harvey, Spellbook), enterprise search (Glean, Hebbia), clinical documentation (Nabla, Suki, Abridge), and contact-center (Observe.AI, Uniphore).
Why: these products sell top-down into large, often regulated organisations, where price scales with headcount and deployment and every deal is negotiated. Gating fits that high-touch motion. It's the buying process — not simply being 'enterprise' — that drives the decision to hide pricing; plenty of enterprise tools still post a public number, and clinical PLG plays for individual clinicians (Heidi, Freed) publish openly even inside a gated vertical.
How it works
Evidence over time
27 supporting · 10 counter — hover or tap a point for detail, click to jump to the row.
Evidence
| Company | Date | What happened |
|---|---|---|
| Harvey | May 2026 | No public pricing surface at all — fully gated, sales-led per-seat quotes (legal vertical). |
| Glean | May 2026 | Pricing confirmed gated; Enterprise Flex seats + pooled FlexCredits documented only via sales (enterprise search). |
| Jasper | May 2026 | Pro seat price published but Business custom-quoted; transparency effectively gated (marketing vertical). |
| Spellbook | Jun 2026 | Legal AI contract tool — fully gated, sales-led, no public dollar figures (legal vertical). Pricing page shows 'Book a Demo' and '$99 for the first month' CLE promo only. |
| Lago | Jun 2026 | Billing-infra. Free self-hosted edition but managed Business/Enterprise tiers are quote-only. |
| Metronome | Jun 2026 | Billing-infra. Free Starter exists, paid plans sales-quoted. Acquired by Stripe Jan 2026. |
| Orb | Jun 2026 | Billing-infra. Sales-led custom pricing, no public rate card. |
| Parloa | Jun 2026 | Contact-center AMP platform — no price table anywhere, /pricing 404s; third-party reviews put the minimum contract near $300K/yr and ACV above $350K (contact-center vertical). |
| PolyAI | Jun 2026 | Sales-only voice assistants for contact centers; the pricing page's first qualifying question is annual call volume, topping out at 'more than 1,000,000' calls (contact-center vertical). |
| Uniphore | Jun 2026 | Enterprise agentic-AI platform at ~$2.5B valuation whose /pricing URL returns a 404 — fully quote-only (contact-center vertical). |
| Observe.AI | Jun 2026 | /pricing returns 404; third-party guides report a ~100-seat minimum and mandatory annual commitment (contact-center vertical). |
| Mercor | Jun 2026 | Human-data marketplace at a claimed $1B+ run rate; every buyer surface terminates in a contact form while the contractor-side pay rate ($141/hr average) is public — one-sided transparency. |
| Robin AI | Jun 2026 | Legal AI. Was Free + $100/user/mo Pro; now gated to sales-only. The gating happened post-growth — a directional move toward opacity. |
| Legora | Jun 2026 | Legal AI — sales-only, seats-only pricing. Confirms clinical/legal vertical gating pattern. |
| Nabla | Jun 2026 | Clinical ambient documentation — gated, seat-based. Third clinical-vertical gated company in corpus. |
| Suki AI | Jun 2026 | Clinical AI voice assistant — sales-only, seat/active-user pricing. Confirms clinical AI as a gated cluster. |
| Observe.AI | Jun 2026 | Contact-center AI — sales-only, per-agent + platform fee. Gated enterprise contact-center AI. |
| Uniphore | Jun 2026 | Contact-center conversational AI — sales-only with per-interaction and per-seat components. Gated enterprise. |
| Artisan | Jul 2026 | Second withdrawal of public pricing, and the most complete de-instrumentation in the corpus. Removed: Free ($0/300 credits), Intern ($250/mo, 12K credits), Employee ($600/mo, 30K credits), the credit estimator, per-action credit costs, the monthly/annual toggle, the plan-comparison table, the 10,000-credit ($300) free trial and self-serve checkout. Replaced by three quote-only tiers sized in leads contacted per month — Team (~2,500), Scale (~6,000, 'Most popular'), Enterprise — each labelled 'Pricing scoped on your plan · Talk to sales.' |
| Inflection AI | Jul 2026 | Gating taken to its limit: inflection.ai/enterprise — the company's only revenue surface, previously a 'Get in touch' page for owned on-prem Inflection 3.0 deployments on Intel Gaudi 3 — now returns a branded 404. Enterprise is absent from nav and footer, developers.inflection.ai no longer resolves in DNS, and no price, pricing page, contact-sales link or quote-request form exists anywhere on the site. Discovered in the same capture that launched Inflection AI Labs. |
| Flexprice | Jul 2026 | Partial gating at a billing-infra vendor: the $1,000/mo Scale tier's CTA changed from self-serve 'Get Started' to 'Contact Us' (sales-assisted), and the cloud Free plan's 1-month validity was replaced by a cap of up to $100K cumulative billing revenue. The published dollar figure stayed; the ability to buy it without a conversation did not. |
| Hyperline | Jul 2026 | Billing-infra vendor moving from a fully published formula to mostly-gated bands: $199/mo + 0.6% of billed revenue and $299/mo + 0.7% were replaced by flat Launch at $599/month (up to $2M revenue, 100 contracts) plus custom-quoted Growth ($2M–$20M) and Scale (above $20M). The 10-invoices-free, no-credit-card trial was removed entirely. |
| m3ter | Jul 2026 | Gated pricing survives acquisition unchanged. Salesforce's investment converted to ownership ('m3ter is now part of Salesforce'), and the /pricing page is word-for-word identical: 'Build your pricing in 4 simple steps' — core platform fee, add-ons, support package, implementation services — with zero dollar amounts and a 'Talk to sales' CTA. |
| Mercor | Jun 2026 | One-sided transparency degrading in place: the Experts hero metric was relabelled from 'average pay $141/hr' to 'average contracted rate $80/hr' — a 43% drop delivered by redefining the metric rather than by a price change — while roles created rose 258.1K → 287.1K and every buyer-side pricing surface remained a contact form. |
| Groq | Aug 2026 | De-instrumentation inside the corpus's LOWEST-gating cohort, and the transparency field cannot see it. groq.com/pricing now 308-redirects to the homepage, which carries no pricing content and no Pricing nav link; the homepage instead leads with an 'ANNOUNCING OUR $650 MILLION FUNDRAISE' banner. No published rate moved — Llama 3.1 8B $0.05/$0.08, Llama 3.3 70B $0.59/$0.79, GPT OSS 120B $0.15/$0.60, GPT OSS 20B $0.075/$0.30, Whisper $0.111/$0.04 per hour, Orpheus $22/$40 per 1M characters are all identical to the 2026-07-21 capture and still live in the GroqCloud docs model catalog. The real loss is narrower and worse: three built-in agentic tools (web search, website visits, code execution) now have no live public rate anywhere, because their docs defer to the page that no longer exists. Because the model rates survive in docs, price_transparency stays 'public' — so the corpus rate above understates this. |
| Moonshot AI | Aug 2026 | A published enterprise price pulled back behind sales, at a lab in an IPO process. The Kimi Enterprise (Business) track on kimi.com/zh-cn/business carried a public $599/yr/seat price with a disclosed feature list (enterprise data privacy, 2x monthly Agent credits, early access to Agent Swarm and Kimi Claw, Kimi Code 5x credits, dedicated support) that was live through July 30. It is gone: no price, no feature list, visitors routed straight to [email protected]. The Kimi assistant subscription ladder ($0 to $199/mo) and the per-token developer API (Kimi K3, the K2 family, moonshot-v1) are unchanged and remain fully public, so this is a single track moving upmarket rather than a site-wide gate. |
| Netlify | Aug 2026 | The smallest possible gating move, and a common one: a disclosed floor deleted. Between the 2026-07-29 and 2026-08-04 captures Netlify dropped the 'starts at $500/month' caption beneath its Enterprise tier's 'Custom' label, leaving only 'Custom' with no dollar figure anywhere in the card. Nothing else moved — Free $0, Personal $9/month, Pro $20/month base with 3,000 credits, the usage-credit rates (production deploys 15 credits, compute 10 credits/GB-hour, bandwidth 20 credits/GB, web requests 2 credits/10k) and the 180-credits-per-$1 AI-inference conversion were all verified unchanged. A visible floor is the last published number most Enterprise tiers carry, and it is the cheapest one to remove. |
Counterexamples
- Perplexity AI · Aug 2026 — De-gating and re-gating in one release, which is why this trend's direction stays two-way. Perplexity's enterprise pricing page had returned a Cloudflare bot-verification challenge to every check since at least 2026-07-21, forcing buyers onto help-centre articles. On 2026-08-14 it came back online with per-seat prices published — Enterprise Pro $34/seat/month and Enterprise Max $271/seat/month, both shown as annual-billed rates, consistent within rounding with the prior $400/seat/yr and $3,250/seat/yr. But the same redesign moved SCIM, audit logs, role-based access control, data-retention configurability and a separate enterprise domain to 'Contract required' on BOTH tiers, replacing the old 50-seat threshold gate, and added a third card ('Additional security and governance', custom pricing) bundling those same features with volume discounts and forward-deployed support. The price became visible; the governance layer became negotiable.
- Higgsfield · Aug 2026 — Opacity by rendering failure, resolved — and it had been hiding a full reprice. Every automated capture of Higgsfield's client-rendered pricing page from 2026-06-05 through 2026-08-11 returned either a loading skeleton or an explicit 'Unable to load pricing plans' error, so its tiers were only ever triangulated from third-party roundups. On 2026-08-14 the page rendered, and the ladder had changed: Free/Starter $15/Plus $49/Ultra $129/Business ~$89-seat is replaced by Basic $9 (120 credits), Pro $29 ($23 annual, 600-900 credits via slider), Max $79 ($59 annual, 1,800-5,400 via slider) for individuals, and Team $79/seat ($65 annual, 2-9 seats) and Scale $215/seat ($150 annual, 5-15 seats) for teams, with no $0 Free card on the grid. A vendor can be effectively gated without intending to be, and this corpus counted it as public throughout.
- Nomic · Aug 2026 — De-gating at the hardest starting point in the set: a flagship sold only behind a seat minimum. Nomic's AEC Platform had listed just Business ($40/user/month, 25-seat minimum, an explicit $1,000/month platform commitment) and custom Enterprise since November 2025, with a single 'Book a Demo' CTA. The page now carries four tiers, adding Free (no credit card, limited agent requests) and Individual ($20/month, self-serve, all product surfaces except the Agent API, AI usage via on-demand top-ups), and the standalone $1,000/mo platform-commitment line is gone. A sales-gated vertical product publishing a self-serve entry price is the direction this trend says is rare, and it happened.
- Apollo · Jul 2026 — De-gating inside the gated cohort. Apollo — one of the 29 gated vertical-SaaS companies — replaced a rate-free 'API Pricing' page and a login-gated About Credits page with a public per-endpoint credit table: people enrichment 1–9 credits (1 for demographics/email, +8 if a mobile phone is returned), email waterfall 1–4 (20+ in some configs), phone waterfall 8–25 (45+ in some configs), 1 credit per organization/person/page, and 0 credits for create/update/list/manage endpoints. Proof that gating is reversible at a sales-led vendor.
- Chargebee · Jul 2026 — A billing-infra vendor publishing MORE, against the segment's 53% fintech gating rate: four product tabs (Billing, CPQ, RevRec, Growth) with Billing Starter at $0 free to the first $250K of cumulative billing then 0.75%, Performance at $7,188/yr for up to $100K billing/mo, a free CPQ Lite for 50 quotes, and only Enterprise quoted. Retention and Receivables were retired rather than gated.
- Intercom · Mar 2024 — Customer-service vertical, yet publishes outcome/resolution pricing openly ($0.99 per Fin resolution).
- Perplexity AI · May 2025 — Enterprise vertical-ish, but ships a public enterprise pricing page.
- Rox · Jun 2026 — Outcome-based agentic AI for sales — publishes all pricing publicly (Free/Core $50/Enterprise).
- Heidi Health · Jun 2026 — Clinical AI for GPs with public pricing (Free / $29 / $59 / Enterprise) — proves the vertical-gating rule has exceptions even in healthcare when the buyer is an individual clinician rather than a hospital system.
- Freed · Jun 2026 — AI medical scribe with public per-seat pricing — a clinical-AI product that uses PLG and publishes prices, showing individual-clinician TAM enables public pricing even in healthcare.
Trivia
-
The wave-27 voice intake (June 2026) produced the corpus's starkest same-category transparency split: contact-center platforms Parloa (~$300K reported minimum), PolyAI, Uniphore, and Observe.AI all 404 or sales-gate their /pricing paths, while the voice-infra vendors one layer down — Retell ($0.055/min), Vapi ($0.05/min), Synthflow — publish per-minute rates to the cent. Same technology, opposite postures, separated only by who signs the contract.
-
The overall corpus gating rate rose from 7% (at 43 companies) to 16% (at 97 companies) to ~20% (31 of 158) not because vertical-knowledge AI became more opaque, but because the expanded corpus added more billing-infrastructure and outbound sales-tech vendors — which gate far above the baseline. Without the billing-infra segment, the gating rate would have risen much less, meaning the headline trend is driven by a product-segment composition shift, not a broad move toward opacity.
-
Robin AI's move from a published Free + $100/user/month Pro page to full sales-only gating is an early documented case in the corpus of a vendor reversing from public pricing to a gated model after growth — the 2026 pricing-page-withdrawal wave later produced several more (Dropzone, Ada, Gladly); most other gating cases were gated from launch. The direction-change suggests that as deal sizes increase in legal AI, the transparency cost (price anchoring in negotiations) outweighs the acquisition benefit (self-serve discovery).
-
Intercom is the sharpest counterexample to the "vertical-knowledge AI gates pricing" thesis: a customer-service vertical product that publishes its outcome price ($0.99 per Fin resolution) openly and self-serve. Its willingness to publish a per-outcome rate while legal and enterprise-search verticals stay gated reflects the difference between a product with a definable, comparable unit (a resolved ticket) and one whose value depends heavily on the buyer's use case and headcount.
-
The corpus gating RATE fell while the count rose. Gated-or-sales-only went from 100 of 338 companies (30%) to 103 of 353 (29%) between 2026-06-10 and 2026-07-30 — three more gated companies against fifteen more companies overall. Every prior review of this trend recorded a rising rate; this is the first that records a falling one, which retires "opacity is spreading" as a corpus-level claim.
-
Gating varies 7.6× by product segment, and the extremes are more informative than the average. customer-service gates 10/19 (53%) and fintech 9/17 (53%); infra-cloud gates 3/42 (7%). A vendor selling GPU-hours by the second against a public spot market cannot gate and survive; a vendor selling a contact-centre transformation has nothing a buyer can price-compare.
-
Artisan pulled its public prices for the SECOND time on 2026-07-14 — and took the measuring instruments with them. Gone in one change: Free ($0, 300 credits), Intern ($250/mo, 12K credits), Employee ($600/mo, 30K credits), the credit estimator, the per-action credit costs, the monthly/annual toggle, the plan-comparison table, the 10,000-credit free trial, and self-serve checkout. What replaced them: three tiers sized in "leads contacted per month", each marked "Pricing scoped on your plan · Talk to sales."
-
Inflection AI went past gating to absence. On 2026-07-22 inflection.ai/enterprise — the company's sole revenue surface — began returning its branded 404, Enterprise vanished from nav and footer, and developers.inflection.ai stopped resolving in DNS. There is now no price, no pricing page, and no contact-sales or quote-request link anywhere on the site: a corpus company with zero commercial surface, discovered in the same capture that announced its new Labs research product.
-
Mercor's one public number moved 43% by redefinition. On 2026-06-30 the hero metric on its Experts page changed from "average pay $141/hr" to "average contracted rate $80/hr" — the same marketplace, a different denominator — while roles created rose from 258.1K to 287.1K and every buyer-side surface remained a contact form. When a vendor publishes exactly one figure, that figure is positioning, not pricing.
For buyers
In gated verticals, expect per-seat quotes scaled to headcount plus a pooled-credit overage (Glean's Flex model), or PEPM contracts in clinical (Nabla, Suki). Benchmark against the public outcome-priced alternative where one exists (Intercom's $0.99/resolution) or the rare public vertical peer (Heidi, Freed) to keep the sales conversation honest.
For vendors
Gating is a deliberate motion: it needs a sales team, quote-to-cash tooling, and a story for why the price isn't public. It correlates with seat-based, regulated, top-down buying — if you sell self-serve to developers or individual clinicians, a public 'starting at' number is increasingly table stakes.
Outlook — what to watch
Gating is a minority but climbing — from ~20% to ~30% as the corpus added more clinical, contact-center, and legal AI, the cohorts that gate hardest. Vertical-knowledge AI is now the sharpest cluster (44% of vertical-SaaS), but the counter-pressure is real: as AI procurement professionalises and individual-clinician TAM opens up, expect more public 'starting at' anchors. The thesis would weaken if the gating rate stops tracking vertical-SaaS composition, or if a marquee gated vertical name (Harvey, Glean) posts a public number moving down-market.
Bottom line
100 of 338 corpus companies gate or hide pricing (~30%) — concentrated in vertical-knowledge AI (44% of vertical-SaaS: legal, clinical, enterprise search, contact-center), billing infrastructure (11 gated), and outbound sales-tech. Verticality alone doesn't force gating; the top-down buying motion does — which is why individual-clinician PLG plays in the same verticals still publish.
FAQ
Why do some AI companies hide their pricing?
Gated pricing fits high-touch, seat-based enterprise sales where price scales with headcount and deployment. In the corpus it clusters in vertical-knowledge AI — 27 of 61 vertical-SaaS companies (44%) gate, spanning legal (Harvey, Spellbook), enterprise search (Glean), and clinical (Nabla, Suki) — plus billing infrastructure and outbound sales-tech. Overall, 100 of 338 companies (~30%) gate.
Is gated pricing common in AI?
No — it's the minority. 238 of 338 corpus companies (~70%) post public prices, including nearly every developer-API and creative-app vendor. Gating has risen from ~20% to ~30% as the corpus added more clinical, contact-center, and legal AI, but it still correlates with the top-down buying motion, not with being 'enterprise'.
Which AI verticals gate their pricing most?
Vertical-knowledge AI gates hardest: 27 of 61 vertical-SaaS companies (44%) hide pricing — legal (Harvey, Spellbook, Legora, CoCounsel), enterprise search (Glean, Hebbia), clinical documentation (Nabla, Suki, Abridge, Ambience), and contact-center (Observe.AI, Uniphore, PolyAI). Billing infrastructure gates even harder as a segment, and outbound sales-tech (11x, Nooks, Apollo) is gated too.
How do I evaluate a vendor with no public price?
Expect per-seat quotes plus a pooled-credit overage (or PEPM in clinical), ask for the per-seat rate and minimums, and benchmark against a public outcome-priced alternative (Intercom, $0.99/resolution) or a rare public vertical peer (Heidi, Freed) to keep the negotiation grounded.