Shortlist by channel
What is the best AI customer service agent for phone calls?
No agent is the best AI customer service agent for phone calls, because nobody has tested them against each other. The best-evidenced are PolyAI, Decagon and Yellow.ai: an external source, not the vendor, reports each of them answering customers over phone calls. 20 of the 22 agents reviewed have any published evidence for phone calls; for the rest it is unknown, which means nobody has published it, not that the product cannot do it. Nothing here has been tested by this portal, so treat every name below as a shortlist to validate, not a winner.
An outside source reports it serving this segment
External report5 of 22A customer, analyst or journalist with no stake in the sale describes the product working here. This is the strongest evidence the portal holds, and it is still somebody else's deployment, not yours.
Ordered by customer workflows with evidence: externally reported first, then vendor-documented, A–Z on a tie. Both counts are shown with every agent. A higher place means more has been established, not that the agent performs better.
- Read the review: PolyAI
Customer workflows with evidence
- 4 of 7 externally reported
- 0 of 7 vendor-documented
- Read the review: Decagon
Customer workflows with evidence
- 3 of 7 externally reported
- 4 of 7 vendor-documented
Read the review: Yellow.aiCustomer workflows with evidence
- 3 of 7 externally reported
- 4 of 7 vendor-documented
- Read the review: NiCE Cognigy
Customer workflows with evidence
- 2 of 7 externally reported
- 5 of 7 vendor-documented
Read the review: ForethoughtCustomer workflows with evidence
- 2 of 7 externally reported
- 2 of 7 vendor-documented
The vendor documents this segment, and nobody else has written about it
Vendor-documented9 of 22The vendor says it serves this segment. That is a reasonable place to start a shortlist and it is not confirmation, because no independent source has reported on it either way.
Ordered by customer workflows with evidence: externally reported first, then vendor-documented, A–Z on a tie. Both counts are shown with every agent. A higher place means more has been established, not that the agent performs better.
- Read the review: Talkdesk
Customer workflows with evidence
- 4 of 7 externally reported
- 2 of 7 vendor-documented
- Read the review: NiCE CXone
Customer workflows with evidence
- 3 of 7 externally reported
- 4 of 7 vendor-documented
Read the review: CrescendoCustomer workflows with evidence
- 2 of 7 externally reported
- 5 of 7 vendor-documented
- Read the review: Sierra
Customer workflows with evidence
- 2 of 7 externally reported
- 5 of 7 vendor-documented
Read the review: AgentforceCustomer workflows with evidence
- 2 of 7 externally reported
- 4 of 7 vendor-documented
- Read the review: LivePerson
Customer workflows with evidence
- 2 of 7 externally reported
- 4 of 7 vendor-documented
Tidio
Lyro
Vendor-documentedWhat puts it here
Documented on getlyro.ai, Tidio's standalone Lyro site, through an ElevenLabs phone integration; Tidio's own help center lists no voice channel for Lyro inside Tidio.
Read the review: LyroCustomer workflows with evidence
- 2 of 7 externally reported
- 3 of 7 vendor-documented
- Read the review: Five9
Customer workflows with evidence
- 2 of 7 externally reported
- 1 of 7 vendor-documented
Read the review: SprinklrCustomer workflows with evidence
- 1 of 7 externally reported
- 5 of 7 vendor-documented
A limitation is documented for this segment
Caveat documented6 of 22A cited source describes a limit or a bad experience here. It sits below vendor documentation because it is not evidence of fit, and above the agents nobody has written about because somebody at least looked. Read the condition: it may not apply to you at all.
Ordered by customer workflows with evidence: externally reported first, then vendor-documented, A–Z on a tie. Both counts are shown with every agent. A higher place means more has been established, not that the agent performs better.
- Read the review: Ada
Customer workflows with evidence
- 3 of 7 externally reported
- 3 of 7 vendor-documented

Kore.ai
Kore.ai
Caveat documentedWhat puts it here
The voice failure reference is a single G2 reviewer, who otherwise reported no major dislikes, noting occasional latency of a few seconds in voice interactions.
Read the review: Kore.aiCustomer workflows with evidence
- 1 of 7 externally reported
- 5 of 7 vendor-documented
- Read the review: Zendesk AI Agents
Customer workflows with evidence
- 1 of 7 externally reported
- 4 of 7 vendor-documented
Read the review: Genesys Cloud CXCustomer workflows with evidence
- 1 of 7 externally reported
- 3 of 7 vendor-documented
- Read the review: Fin
Customer workflows with evidence
- 0 of 7 externally reported
- 5 of 7 vendor-documented

HubSpot
Breeze Customer Agent
Caveat documentedWhat puts it here
Voice is a public beta that answers inbound calls only, and the agent's actions settings do not apply to calls.
Read the review: Breeze Customer AgentCustomer workflows with evidence
- 0 of 7 externally reported
- 4 of 7 vendor-documented
Nothing published for this segment
Not established2 of 22No reviewed source places these agents here, by the vendor's own documentation or otherwise. That is a gap in the public record, not a finding against the products, and several of them may suit you.
Ordered by customer workflows with evidence: externally reported first, then vendor-documented, A–Z on a tie. Both counts are shown with every agent. A higher place means more has been established, not that the agent performs better.
Read the review: Freddy AI AgentCustomer workflows with evidence
- 0 of 7 externally reported
- 5 of 7 vendor-documented
- Read the review: Gorgias AI Agent
Customer workflows with evidence
- 0 of 7 externally reported
- 5 of 7 vendor-documented
Every claim behind these groups
Each agent above with evidence for phone calls, and the claims that placed it, with the source type and the date it was checked.
PolyAI
External reportPolyAI is voice-first: its documentation index describes "enterprise voice and chat agents" and its pricing is per minute of voice-agent use, with secondary documented channels covering web chat, SMS and RCS, email, and iOS/Android SDKs (including in-app voice calling).Vendor-reported
- Source
- PolyAI Docs — documentation index (llms.txt)
- Checked
- History
- First published · Last changed
- Correct this
- Request a correction with a public source. Claim ID
polyai-cmp-channels
A director of contact-centre operations wrote on G2 that PolyAI deployed a voice assistant for non-revenue calls "in 4 weeks" and "From day 1 it handled 87% of these calls", and that they were "exploring other opportunities to allow full hotel reservations bookings".Independent report
- Source
- PolyAI reviews on G2, enterprise reviewer
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
polyai-g2-hotel-containment
Forbes reported in December 2025 that PolyAI is "managing customer calls about power outages for PG&E, California’s largest energy company".Independent report
- Source
- PolyAI Raises $86 Million As Fight To Answer Calls With AI Heats Up (Forbes)
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
polyai-press-forbes-pge
NS&I (National Savings and Investments, under HM Treasury) states in its UK government algorithmic-transparency record that PolyAI is in production "Answering general product and service queries from NS&I customers by phone", handling over 100,000 interactions a month, and that "100% of inbound customer calls are now answered by PolyAI".Independent report
- Source
- NS&I: PolyAI (UK algorithmic transparency record, GOV.UK)
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
polyai-press-nsi-product-queries
The UK Ministry of Justice's algorithmic-transparency record states that a PolyAI voice assistant is in production at the Prison Enquiry Centre for five prisons, answering "frequently asked process questions" from prisoners' friends and families, and that when no FAQ is identified, a safeguarding issue is detected or the caller asks for an agent, "the call is transferred to the next available human agent".Independent report
- Source
- MoJ: PolyAI / Luna (UK algorithmic transparency record, GOV.UK)
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
polyai-press-moj-prison-enquiries
Decagon
External reportDecagon’s homepage lists chat, email and voice. SMS appears in its Curology customer story; confirm whether that channel is available for your proposed deployment.Vendor-reported
- Source
- Decagon homepage and linked customer examples
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
decagon-channels
A G2 reviewer at a capital-markets firm wrote on July 31, 2026 that Decagon handles “automated calls and automated emails” for them. The review does not describe the setup.Independent report
- Source
- abhay pratap singh c., Decagon review on G2
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
decagon-g2-calls-emails
Yellow.ai
External reportAn Escalate to Agent tool places chat conversations in Inbox queues with configurable priority and assignment and delivers a conversation summary to the human agent, plus a Transfer Call tool for PSTN forwarding or SIP transfer on voice.Vendor-reported
- Source
- Escalation (Yellow.ai docs)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
yellow-ai-cmp-handoff
A senior software engineer reviewing yellow.ai on G2 writes that “phone and email configuration are straightforward to implement” and that “Primarily I'm working in the Hospitality and Airlines Business”.Independent report
- Source
- yellow.ai reviews on G2
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
yellow-ai-g2-voice-email
NiCE Cognigy
External reportDocumented endpoints cover web chat, telephony via Cognigy Voice Gateway, WhatsApp, Facebook Messenger, Microsoft Teams, Slack, LINE Messenger, Twilio and Twilio SMS, Intercom, Userlike, Zendesk Sunshine Conversations, Genesys, 8x8, NiCE CXone, RingCentral Engage, Zoom, Azure Bot Services, plus REST, Socket and Webhook endpoints.Vendor-reported
- Source
- Endpoint Reference - Overview | Cognigy Documentation
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
nice-cognigy-cmp-channels
A G2 reviewer in IT services describes using Cognigy.AI for "quick implementation of message and voice conversations", building virtual assistants integrated with business platforms that respond "to voice commands and messages in different channels with little or no programming."Independent report
- Source
- Cognigy.AI reviews on G2
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
nice-cognigy-g2-fit-mid-market-2
Forethought
External reportThe vendor lists chat, email, voice, SMS, Slack, mobile apps and a headless API as supported channels.Vendor-reported
- Source
- Forethought Solve
- Checked
- History
- First published · Last changed
- Correct this
- Request a correction with a public source. Claim ID
forethought-channels
SiliconANGLE reports that the platform supports interactions across channels such as email, chat, messaging and voice systems.Independent report
- Source
- Zendesk announces deal to acquire customer support AI startup Forethought
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
forethought-press-channels
Talkdesk
Vendor-documentedTalkdesk says its AI agents connect interactions across voice, digital, email and web, and describes Autopilot as resolving requests across voice and digital channels.Vendor-reported
- Source
- Talkdesk AI Agents / Multi-Agent Orchestration
- Checked
- History
- First published · Last changed
- Correct this
- Request a correction with a public source. Claim ID
talkdesk-cmp-channels
NiCE CXone
Vendor-documentedNiCE states its self-service AI agents deliver natural language experiences that are equally fluent over digital or voice channels, with multimodal capabilities blending voice, text, multimedia and device capabilities.Vendor-reported
- Source
- AI Agents for Self-Service
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
nice-cxone-cmp-channels
Crescendo
Vendor-documentedCrescendo states its Concierge agent handles "Voice, chat, email, SMS, WhatsApp" from a single AI layer, with a website chat widget, mobile apps and API access also documented.Vendor-reported
- Source
- Crescendo — Concierge Agent
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
crescendo-cmp-channels
Sierra
Vendor-documentedVendor-listed channels include phone, website and app chat, WhatsApp, Apple Business Chat, SMS and email. Live Assist supports staff, and Sierra also documents publishing selected agent journeys to ChatGPT.Vendor-reported
- Source
- Channels (Sierra)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
sierra-channels
Agentforce
Vendor-documentedSalesforce describes messaging, web, mobile and voice service experiences. Channel availability and setup should be checked for the specific Service Agent offering.Vendor-reported
- Source
- AI customer service agents (Salesforce)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
agentforce-channels
LivePerson
Vendor-documentedLivePerson states its AI agents can be built and launched with LLMs 'across messaging and voice', and its Conversational Cloud page names SMS, WhatsApp and voice among supported channels.Vendor-reported
- Source
- Generative AI Tools — LivePerson
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
liveperson-cmp-channels
Lyro
Vendor-documentedDocumented on getlyro.ai, Tidio's standalone Lyro site, through an ElevenLabs phone integration; Tidio's own help center lists no voice channel for Lyro inside Tidio.
Tidio's standalone Lyro site says any Lyro agent can be wired into a phone line through ElevenLabs, using the same knowledge and skills, with the full transcript handed back to the help desk after every call.Vendor-reported
- Source
- The AI Customer Service Agent That Grows Revenue - Lyro (getlyro.ai)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
lyro-doc-voice
Five9
Vendor-documentedFive9 markets both "Voice AI Agents" for automating routine customer interactions and "Digital AI Agents" providing personalized self-service via chatbots and social messaging.Vendor-reported
- Source
- Conversational, Generative and Agentic AI
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
five9-channels
Sprinklr
Vendor-documentedSprinklr states AI agents operate across "chat, social, voice, messaging, email, and more", moving between voice, chat, email and social while retaining conversation context.Vendor-reported
- Source
- Sprinklr AI Agents
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
sprinklr-channels
Ada
Caveat documentedChat, voice, email and social messaging are documented. Voice and social connections use supported external platforms; confirm the intended channel integration.Vendor-reported
- Source
- Key concepts (Ada docs)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
ada-channels
Ada’s language support differs by channel, and its management dashboard is English-only. A language available in email should not be assumed available for voice.Vendor-reported
- Source
- About multilingual support (Ada docs)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
ada-voice-language-subset
Kore.ai
Caveat documentedThe voice failure reference is a single G2 reviewer, who otherwise reported no major dislikes, noting occasional latency of a few seconds in voice interactions.
Kore.ai's channel documentation lists digital channels (Slack, LINE, Microsoft Teams, WhatsApp, Messenger, Twilio SMS, Telegram, Zendesk, Instagram, Genesys, Genesys Open Messaging, Email), voice channels (Realtime LLM Voice, Pipeline Voice, VXML IVR, Genesys Audio Connector, AudioCodes), SDK channels (Web SDK, API, Mobile SDK), webhooks, and the AG-UI and Agent-to-Agent protocols.Vendor-reported
- Source
- Channels - Kore.ai Agent Platform documentation
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
kore-ai-cmp-channels
A G2 reviewer of Kore.AI (vendor-level listing) who built an agent for L1 support wrote that "Using the bot we are able to provide voice support also".Independent report
- Source
- Kore.AI reviews on G2
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
kore-ai-g2-voice-support
A G2 reviewer of Kore.AI (vendor-level listing) who said they had no major dislikes noted that "during some enterprise implementations I’ve occasionally noticed a few seconds of latency in the Agent Platform, which can reduce the responsiveness of voice-based interactions".Independent report
- Source
- Kore.AI reviews on G2
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
kore-ai-g2-voice-latency
Zendesk AI Agents
Caveat documentedZendesk advertises web, mobile, social, email, and voice. Check channel-specific documentation and availability for the selected agent.Vendor-reported
- Source
- Zendesk AI Agents product overview
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
zendesk-cmp-channels
Zendesk documents a rare known limitation of voice AI agents: the agent can occasionally produce muffled, distorted or otherwise unusual-sounding speech, which Zendesk cannot resolve during a live call. Its recommended mitigation is choosing a default voice, such as Marin or Cedar, instead of a language-specific voice before deployment.Vendor-reported
- Source
- Voice AI agent pronunciation issues (Zendesk help)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
zendesk-ai-agents-doc-voice-pronunciation
Genesys Cloud CX
Caveat documentedGenesys documents that bots built in Architect can be integrated into call, chat and message flows, and markets AI agents across voice, text, web, social and WhatsApp.Vendor-reported
- Source
- About Genesys Dialog Engine Bot Flows - Genesys Cloud Resource Center
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
genesys-cmp-channels
In a YouTube tutorial recorded in his own test lab, Seth On Call built a Genesys bot flow with intents, slots and a knowledge base, added it to an inbound call flow and showed it recognising the caller's intent and collecting a shopper ID and coupon code by voice, after which the call flow played his scripted confirmation: "We have applied the coupon to your account. Thanks for calling in."Independent report
- Source
- How to Build Bot Flows in Genesys Cloud Architect (Step-by-Step Tutorial) (YouTube, Seth On Call)
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
genesys-cloud-cx-yt-voice-demo
Replying on Genesys's community forum to reports that voice bots treat background sounds as caller utterances, Genesys's Principal Product Manager for Virtual Agent wrote that the speech models "do handle background noise, its just not by the same type of microphone sensitivity, so clearly there are still a few gaps".Vendor-reported
- Source
- Voice bots - background noise - Genesys Cloud Community
- Published
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
genesys-cloud-cx-doc-voice-noise-gap
Fin
Caveat documentedIntercom documents Fin Voice answering phone calls from the support knowledge base and handing calls that need a person to a teammate, a workflow, a callback or another number; it lists 28 languages for Fin Voice.Vendor-reported
- Source
- Deploy Fin Voice (Intercom Help)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
fin-doc-voice
Intercom says Fin Voice "is currently limited to select customers partnered with our Sales team", and that a Fin Voice call ends or hands off after a maximum of 29.5 minutes.Vendor-reported
- Source
- Deploy Fin Voice (Intercom Help)
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
fin-doc-voice-availability
Breeze Customer Agent
Caveat documentedVoice is a public beta that answers inbound calls only, and the agent's actions settings do not apply to calls.
HubSpot lists live chat, email, forms, WhatsApp, Facebook Messenger, and inbound voice calls as channels the agent can handle.Vendor-reported
- Source
- AI Customer Service Agent — HubSpot
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
hubspot-breeze-cmp-channels
HubSpot documents Customer Agent Voice, in public beta, as handling inbound calls only, with outbound calling not supported yet, and states that the agent's actions settings don't apply to calling channels (voice handoff is managed in the calling channel's settings instead).Vendor-reported
- Source
- Set up customer agent voice — HubSpot Knowledge Base
- Checked
- History
- First published
- Correct this
- Request a correction with a public source. Claim ID
hubspot-breeze-doc-voice-limits
How these groups are formed
- Agents are grouped by what the cited sources establish for this segment, not scored against each other.
- An outside source reporting the product in this segment comes first, then the vendor's own documentation, then a documented limitation, then silence.
- Within a group, the agent with more of the seven customer workflows reported by an outside source comes first, then the one with more vendor-documented workflows; only where both counts tie is the order alphabetical.
- No agent is dropped: the ones with nothing published are named too.
Groups and the order inside them are derived from the cited claims at build time and describe the evidence, not the products. Nothing here has been tested, so a higher group or a higher place inside one means more has been established about an agent, not that it performs better. An agent can be well suited to you and sit in the last group simply because nobody has written about it. Published review-site ratings do not order these groups either: where a record carries one it is cited on the profile, but those ratings score whole products against each other — one vendor's bundled chat widget against another's full help desk — and none of them measures the AI agent on its own.
The methodology page explains why the portal groups evidence rather than scoring products.
Other shortlists
Facts last checked . No agent on this page has been tested by this portal.