# State of AI voice agents, November 2026: What buyers ask AI and what vendors publish

By Samuel Oram-Jones

The state of AI voice agents as of November 2026, in figures: buyers search most for ai receptionist, ai call center, and ai voice agents, Google's AI answers name Retell, ElevenLabs, and Bland most among the vendors counted, and vendors price by seat, by minute, or by outcome. Each figure traces to a capture dated 20 November 2026 or earlier.

Bland publishes this report and is counted by the same method as every other vendor, including where its count is low. Where Bland is absent from an answer, the count says so.

## What do buyers search for when they look for AI voice agents?

Buyers search for AI receptionists, AI call centers, and AI voice agents far more than for any other term in the set: 74,000, 49,500, and 40,500 searches a month in the United States, more than six times the next term. Those are Google Ads averages captured on 20 November 2026, and head-term averages are inflated by a July to October 2026 spike, so a typical month sits below them.

Below the head terms the volumes fall away fast. Of the 28 unbranded terms that returned a volume, 10 exceed 1,000 searches a month and 13 sit at 100 or fewer. Another 29 of the 62 terms measured returned no volume at all. Search volume counts what people type, and says nothing about how large a market is.

Call center terms carry ai call center at 49,500, ai contact center at 4,400, and ai call center software at 3,600. Builder terms carry ai agent builder at 6,600, no code ai agent builder at 5,400, and voice agent builder at 70. Rival-alternative terms are small: synthflow alternative at 390, polyai alternatives at 210, retell ai alternatives at 50, and voiceflow alternative at 20.

Terms that name Bland run from bland ai at 6,600 down to bland ai alternatives at 320, bland ai pricing at 260, bland ai review at 70, and bland norm at 10.

Government and FedRAMP terms barely register: of 8 measured, 7 returned no volume and fedramp contact center returned 10. Report-style terms are small as well, with voice ai market at 10 and state of voice ai 2026 returning none.

## Which vendors do AI answers name for AI voice agent questions?

Across 112 unbranded buyer keywords, Google's AI Overview or AI Mode named Retell most often, in 29 of the 110 keywords that do not already name Retell, followed by ElevenLabs in 27 of 112 and Bland in 25 of 112. Each vendor is counted only on keywords that do not contain its own name, which is why the denominators differ.

| Vendor | Keywords counted | AI Mode | AI Overview | Either |
|---|---|---|---|---|
| Retell | 110 | 18 | 25 | 29 |
| ElevenLabs | 112 | 20 | 19 | 27 |
| Bland | 112 | 17 | 16 | 25 |
| Genesys | 108 | 18 | 12 | 24 |
| Vapi | 112 | 19 | 17 | 24 |
| Synthflow | 111 | 11 | 16 | 21 |
| NICE (CXone) | 110 | 11 | 13 | 18 |
| Zendesk | 112 | 14 | 8 | 18 |
| Talkdesk | 112 | 12 | 8 | 16 |
| Five9 | 110 | 9 | 7 | 14 |
| Voiceflow | 112 | 6 | 11 | 14 |
| Amazon Connect | 111 | 9 | 7 | 13 |
| Twilio | 112 | 9 | 7 | 13 |
| Cognigy | 111 | 4 | 9 | 10 |
| LiveKit | 112 | 5 | 7 | 9 |
| Kore.ai | 111 | 4 | 5 | 8 |
| PolyAI | 107 | 4 | 7 | 7 |
| Decagon | 111 | 2 | 5 | 5 |
| Sierra | 111 | 1 | 4 | 4 |
| Parloa | 112 | 0 | 3 | 3 |

Bland is named in 17 AI Mode answers and 16 AI Overviews, and in at least one of the two for 25 keywords, which leaves 87 keywords where neither answer names Bland. For 12 of the 112, bland.ai ranks in Google's organic top 10. For 83 keywords Bland shows up on none of the three surfaces: organic top 10, AI Overview, or AI Mode.

A name counts whether an answer recommends a vendor or only lists it, and a count says how often a vendor is named in an answer, not how well the answer speaks of it. Salesforce and HubSpot are not tracked as vendors, since they are CRMs that voice agents connect to.

### Call center and enterprise questions

For the 20 call center and enterprise keywords, AI answers named Genesys 12 times, Zendesk 9, Talkdesk 8, Five9 7, Retell 7, NICE (CXone) 6, Vapi 6, and Amazon Connect, Bland, ElevenLabs, Synthflow, and Twilio 5 each.

Genesys, Zendesk, and Talkdesk each sell seat-priced contact center or customer service platforms, so the first three places in this group go to platform vendors rather than to voice agent specialists. PolyAI was named in 2 of the 20.

Bland was named for ai call center, cost of ai call center, enterprise voice ai platform, voice ai for enterprise, and voice ai market 2026, and ranks in Google's organic top 10 for 3 of the 20. Bland's guide to [running an enterprise contact center on AI voice agents](/enterprise-ai-voice-agents) and its [ranked list of contact center vendors](/blog/best-voice-ai-agents-contact-centers) cover these terms for buyers who want the longer read.

### Government questions

Government questions leave most keywords without a vendor name. Of the 22 government keywords, 14 produced answers that name none of the 20 vendors counted. Across all 22, ElevenLabs was named for 5, NICE (CXone) for 3, and Amazon Connect, Bland, and Genesys for 2 each.

Bland was named for fedramp ai voice agent and fedramp voice ai, and both keywords contain the word FedRAMP. Bland ranks in Google's organic top 10 for 1 of the 22. Bland's [FedRAMP page](/fedramp) holds the current status.

### No-code and small-business questions

For the 15 no-code and small-business keywords, Retell was named for 7, Bland for 6, ElevenLabs for 6, Vapi for 6, Synthflow for 6 of 14, and Voiceflow for 5.

Bland was named for ai phone agent small business, ai voice agent for mid size business, no code ai voice agent, no code voice ai platform, synthflow alternative, and voice ai for smb, and ranks in Google's organic top 10 for 1 of the 15.

## What do AI answers say about Bland when they name it?

When AI answers name Bland, 19 of the 25 also use the word outbound. Google's AI Mode answer for best ai voice agent platform 2026 labels Bland "Best for High-Volume Enterprise Outbound", and the AI Overview for no code voice ai platform says "Bland leans more toward outbound calls".

Bland describes itself differently: Inbound and outbound AI phone agents, built and edited without code in Norm or Bland's visual builder. Bland's own FAQ answers the label directly: Is Bland only for outbound calls? No.

For the 23 keywords that name a rival, Bland was named for 2 and was absent on all three surfaces for 20. Two of Bland's newest posts, one on contact center vendors and one on no-code against custom-code agents, both published on 7 November 2026, appear in none of the 120 captures.

AI answers do cite Bland's pages. Bland's own pages are cited for 13 keywords, across 9 distinct pages: the home page, its FedRAMP page, one alternatives page, and 6 blog posts, among them [Bland's comparison of voice AI platforms for business](/blog/best-voice-ai-platforms-for-business).

For a call outcome rather than a count, here is one Bland customer's own result, from a housing finance association that replaced its IVR:

Idaho Housing and Finance Association saves $750,000 a year by replacing its IVR with Bland, which handles 4,000 inbound calls a day and cut average call time by 20%.

> "We save thousands of dollars a day."

Greg Blake, Chief Information Officer, Idaho Housing and Finance Association

That is one customer's report, not a market statistic and not a forecast, so a reader should not assume it applies to another vendor's calls.

## Which sources do AI answers cite?

YouTube is the source AI answers cite most, for 85 of the 112 unbranded keywords, ahead of Reddit at 51 and LinkedIn at 20. A source counts as cited when it appears in the source list of a Google AI Overview or AI Mode answer, and a keyword counts once however many of a site's pages are cited.

Vendor sites and Facebook follow the three social and video sources. Retell's pages are cited for 17 keywords, Facebook for 15, Synthflow for 14, Bland's own pages for 13, Voiceflow for 13, NICE for 12, RingCentral for 12, Aircall for 11, and ElevenLabs for 10. Genesys, Zendesk, FedRAMP Marketplace listings, and one other vendor's site are cited for 9 each, and Talkdesk, Verint, and Dialpad for 8 each.

Vendor home pages are cited on broad terms. Voiceflow's home page is cited for 10 keywords, Synthflow's for 8, Retell's for 7, and Vapi's for 5. Bland's home page is cited for 1.

## What do the pages AI answers cite have in common?

The pages AI answers cite tend to be long and dated, and most carry an FAQ and name customers. Of 42 non-Bland pages that AI answers cited or Google ranked in its top 10 for these keywords, and that the sweep fetched, the median runs 2,776 words in server HTML, 30 carry an FAQ, 17 carry FAQPage schema, and 15 carry a table.

Dates and figures are common as well. Of the same 42, 25 show a 2026 date, 19 carry at least 15 figures, and 25 name customers. Fourteen are vendor home or product pages, cited for broad terms when the title and headline name the category and the buyer.

Ranked and alternatives lists run long. Seven such pages run 2,928 to 7,862 words, and every one has a table and an FAQ. Two other cited pages returned HTTP 403 on 20 November 2026 and are not among the 42. These pages share traits, and a capture cannot say which trait earns the citation.

For buyers weighing options, Bland's [comparison of AI voice agent pricing](/blog/voice-ai-platform-pricing-comparison-2026) and its [guide to no-code against custom-code agents](/blog/no-code-vs-custom-code-voice-ai-agents) go deeper on cost and build choices.

## How do vendors price AI voice agents?

Vendors price AI voice agents three ways: per seat, per minute, and per outcome. A per-seat price counts agents on a platform, a per-minute price counts time on calls, and a per-outcome price counts finished tasks. The captures below were read on 5 November 2026, and each quote is the vendor's own wording.

Five vendors read for this report price per seat. Genesys publishes seat plans at "$75 USD", "$115 USD", and "$155 USD", billed "per user/month billed annually". Talkdesk publishes "Per user, Per month" rates of "$85", "$105", "$165", and "$225" across four tiers. On its UK site, Zendesk's seat prices read "£15 agent/month paid yearly", "£45 agent/month paid yearly", and "£89 agent/month paid yearly".

Five9 quotes prices "per concurrent user", and NICE CXone sells packages "Per agent per month".

Three price per minute. Amazon Connect prices voice at "$0.038 per voice min (standard telephony rates apply)". Retell publishes "$0.07-$0.31 / min for AI Voice Agents", with the model and voice priced as separate lines. PolyAI quotes its managed agent per minute, "Ongoing use of the voice agent is priced on a per-minute basis", and sells a model API at "2¢ Per connected minute" with "Telephony and orchestration Not included".

Bland prices per minute as well: $0.14 a minute on the self-serve Start plan. Enterprise pricing is custom, set with Bland's sales team. Phone is billed separately, on a customer's own carrier or Bland's at pass-through cost. Bland's [pricing page](/pricing) holds the plan rows, and its [guide to usage-based against outcome-based pricing](/blog/usage-vs-outcome-based-pricing) covers the trade-off in more detail.

Three price per outcome. Sierra's blog on outcome-based pricing reads "With outcome-based pricing, Sierra gets paid only when we complete a task for you".

Decagon offers "Per-conversation pricing: A fixed rate for every incoming conversation" and "Per-resolution pricing: A higher fixed rate for each fully resolved conversation, with no charge for escalations."

Zendesk says its AI agents are "included in every Suite and Support plan, with pricing based on the successful outcomes they deliver".

Outcome pricing carries a forecasting catch that Decagon states itself: "Resolutions may vary month-to-month, making it tough to forecast costs compared to usage-based pricing." Across the 10 vendors read, that makes 11 published pricing models, because Zendesk uses two: 5 per seat, 3 per minute, and 3 per outcome.

A buyer who wants to compare these on one line can start from [what a minute of calls costs against staffed seats](/blog/total-cost-of-ownership-ai-voice-agents-vs-call-center-staff).

## How do vendors state their FedRAMP status?

Vendors state FedRAMP status in at least four forms, from a class with a package ID to an authorization level with no class or date. Each line below comes from the vendor's own page or its FedRAMP Marketplace listing, fetched on 20 November 2026, with quotation marks around wording copied as printed. The capture covers 22 vendors and listings, Bland included.

Genesys prints a class and a package ID, "Genesys Cloud is FedRAMP Certified (Class C)" followed by "(Package ID: FR2131766015)". A second Genesys page, dated 24 August 2026, says Genesys Cloud CX is FedRAMP-authorized at the Class C (Moderate) impact level and listed on the FedRAMP Marketplace under the same package ID.

NICE CXone prints a level and a package ID on its government page: FedRAMP Authorization at the Moderate Impact Level, Package ID FR1704655535.

The FedRAMP Marketplace lists ElevenLabs as "FedRAMP Certified As of 9/23/2026", a 20x listing in the Program path at Class A (Pilot). The Marketplace also lists Lumen Cloud Contact Center as certified as of 30 April 2025 and Content Guru Storm Cloud Contact Center as certified as of 20 March 2025.

Other vendors state an authorization without a class or date in the line extracted from their pages. Zoom's page says its FedRAMP environment is authorized at Moderate level, and Veritone's page says Veritone is FedRAMP Authorized to Operate.

Bland states its status as one line: FedRAMP 20x Class C (Moderate), certified 17 November 2026, FedRAMP Marketplace FR2628647242. Every FedRAMP Marketplace listing carries a package ID. Of the vendors' own status sentences captured, two print it: Genesys and NICE CXone.

AI answers captured on 20 November 2026 for three keywords, bland ai fedramp, bland ai government, and fedramp ai voice agent, give Bland a FedRAMP class and certification date that differ from that line. Bland's [trust and security page](/trust-security) and [FedRAMP page](/fedramp) carry the current status, and the FedRAMP Marketplace lists it too.

## How was this report put together?

Each figure traces to a dated capture. The sweep behind the vendor, source, and page counts ran on 20 November 2026 on Google in the United States, in English, on desktop, and recorded the organic results, the AI Overview, and the AI Mode answer for each of 120 buyer keywords. Eight of the 120 contain the word bland and are set aside, which leaves 112 unbranded keywords. No keyword failed.

A vendor counts as named when its name appears in the text of an AI Overview or an AI Mode answer. Twenty vendors were tracked, from Bland, Retell, and Genesys to Sierra and Parloa, and each is credited only on keywords that do not contain its own name. A source counts as cited when it appears in the source list of an AI Overview or AI Mode answer.

The counts are one capture on one day, and AI answers change from one run to the next. About half of the organic rows have a URL inferred from the display link, and AI Mode returns only a few sources per answer, so source counts are floors. The 42 pages in the page analysis were fetched on 20 November 2026 and measured as served in HTML.

Search volumes come from Google Ads data for the United States in English, fetched on 20 November 2026, and 29 of the 62 terms returned none. Rival pricing was read from each vendor's own pages on 5 November 2026, and FedRAMP lines were read from vendor pages and FedRAMP Marketplace listings on 20 November 2026. Bland's own facts come from Bland's published facts.

This report carries no market-size figure, analyst forecast, or revenue estimate, because no captured source supports one. Bland keeps the captured files and the one script that reproduces every count.

## Frequently asked questions

### Which vendors do AI answers name most for AI voice agent questions?

Across 112 unbranded keywords on 20 November 2026, Retell was named in 29 of 110, ElevenLabs in 27 of 112, Bland in 25 of 112, and Genesys in 24 of 108. Counts cover Google's AI Overview and AI Mode answers, and a vendor is not counted on keywords that already name it.

### Does Bland appear in AI answers for call center questions?

Bland was named for 5 of the 20 call center and enterprise keywords, while Genesys was named for 12, Zendesk for 9, and Talkdesk for 8. Bland ranks in Google's organic top 10 for 3 of those 20 keywords.

### Do AI answers call Bland an outbound-only vendor?

AI answers describe Bland as an outbound vendor in some cases: 19 of the 25 answers that name Bland also use the word outbound. Bland's own FAQ puts it this way: Is Bland only for outbound calls? No.

### How do vendors price AI voice agents?

Per seat, per minute, or per outcome. Genesys, Talkdesk, Zendesk, Five9, and NICE CXone price per seat, Amazon Connect, Retell, and PolyAI price per minute, and Sierra, Decagon, and Zendesk's AI agents price per outcome. Bland's rate is $0.14 a minute on the self-serve Start plan. Enterprise pricing is custom, set with Bland's sales team.

### Which sources do AI answers cite most?

YouTube, cited for 85 of 112 unbranded keywords, then Reddit for 51 and LinkedIn for 20. Vendor sites follow, with Retell's pages cited for 17 keywords and Bland's own pages for 13.

### Does this report give a market size?

No. No captured source gives one, so the report counts what buyers search, what AI answers name, and what vendors publish. Search volume measures typing, and a market-size figure would need a source this report does not have.
