PolyAI
Voice agents for big contact centersEnterprise voice AI platform whose agents handle inbound customer calls end to end — authentication, billing, bookings and routing — across dozens of languages.
- Maker
- PolyAI
- Founded
- 2017
- Pricing
- No free tier · from Custom
- Last checked
- August 2026
Verdict
Enterprise voice AI platform whose agents handle inbound customer calls end to end — authentication, billing, bookings and routing — across dozens of languages.
Score breakdown
vs ElevenLabs, the category leaderLarge enterprises in banking, healthcare, travel, utilities or retail handling millions of inbound calls a year that need deflection without degrading CX.
Key features
- Agent Studio for non-technical teams and an agent development kit for engineers
- Proprietary dialog models trained on over a billion enterprise conversations
- Voice and chat channels from the same agent definition
- Multilingual agents with consistent brand voice
- Call routing, authentication, billing and payments, bookings and order management flows
- Real-time analytics with visibility into agent decisions
- Governance guardrails and brand-safety controls
- SOC 2, HIPAA, GDPR and PCI DSS compliance
Strengths
- +Among the most proven voice platforms at genuine contact-center scale (millions of calls)
- +Purpose-built dialog models trained on enterprise conversations, not a generic LLM wrapper
- +Strong compliance posture out of the box: SOC 2, HIPAA, GDPR and PCI DSS
- +Deep multilingual coverage with consistent brand voice and governance guardrails
- +Managed service model includes ongoing tuning, maintenance and 24/7 support in the per-minute rate
Weaknesses
- −No public pricing and no self-serve tier — everything is a custom quote after a sales cycle
- −Third-party estimates put entry contracts around six figures a year, ruling out SMBs
- −Deployments are project-sized: integration with CRM, telephony and back-office systems takes real time
- −ROI case depends on very high call volume; smaller call centers will not clear the fixed cost
Overview
Overview
PolyAI is a London-based enterprise voice AI company founded in 2017 by Nikola Mrkšić, Tsung-Hsien Wen and Pei-Hao Su, who met at the University of Cambridge and previously worked on conversational systems at Apple, Google and Meta. The product is a platform for building voice agents that answer inbound customer calls and complete the task — authenticating a caller, taking a payment, changing a booking, routing to the right team — rather than reading a menu and giving up.
The technical position is that generic LLMs are not enough for telephony. PolyAI runs proprietary dialog models trained on over a billion enterprise conversations, tuned for the specific failure modes of phone calls: interruptions, accents, background noise, partial information and callers who change their mind mid-sentence. Agents are built either in Agent Studio, aimed at non-technical CX teams, or through a developer kit, and the same definition serves voice and chat. Compliance coverage — SOC 2, HIPAA, GDPR, PCI DSS — is table stakes for the sectors it sells into, and named customers include PG&E, Unicredit, Marriott and FedEx.
Commercially it is unambiguously enterprise. There is no free tier, no self-serve signup and no published rate card: billing is per minute, quoted against call volume, peak concurrency, language count and integration depth, with maintenance and 24/7 support folded into the rate. Third-party estimates put entry contracts around $150K a year. The company raised an $86M Series D in December 2025 co-led by Georgian, Hedosophia and Khosla Ventures, taking total funding past $200M. For a mid-sized support team this is the wrong tool — the fixed cost and integration effort only make sense above a very high call volume.
Key Benefits
- Proven at scale: Deployments handle millions of real customer calls, a track record few voice vendors can match.
- Task completion, not deflection theater: Agents authenticate callers and transact against back-office systems instead of collecting intent and transferring.
- Regulated-industry ready: Compliance certifications and governance guardrails suit banking, healthcare and payments workloads.
- Multilingual by design: One agent can serve many languages while holding a consistent brand voice.
- Managed, not shipped: Ongoing tuning and monitoring are part of the contract rather than an add-on the buyer staffs internally.
Use Cases
- Inbound customer service automation — Handles billing questions, account changes and troubleshooting for utilities, telcos and financial services without a human agent.
- Bookings and reservations — Takes, changes and cancels reservations for hospitality and travel brands, including during demand spikes that would otherwise queue.
- Caller authentication and routing — Verifies identity and passes fully contextualized calls to the correct human team, cutting handle time on transfers.
- Out-of-hours coverage — Provides consistent 24/7 answering in multiple languages where staffing overnight shifts is impractical.
Pricing
Checked August 2026 · USD- Per-minute billing quoted on call volume and concurrency
- Rate includes maintenance, tuning and 24/7 support
- Pricing scales with languages deployed and systems integrated
- Third-party estimates place entry contracts near six figures annually