×
PolyAI
Enterprise Voice Agents
Last verified August 17, 2026 · Updated daily
What PolyAI is building
The phone call is the most important customer service channel that technology has consistently failed to improve. Despite decades of investment in Interactive Voice Response (IVR) systems, chatbots, and call centre automation, the experience of calling a large enterprise for help remains, for most people, a reliable source of frustration: long hold times, scripted systems that can't handle anything outside a narrow decision tree, and the eventual transfer to a human agent who has to be told everything from the beginning. The problem is not that enterprises haven't tried to automate it. The problem is that every previous generation of voice automation was built on rules – rigid, hand-crafted logic that broke the moment a caller deviated from the expected script. PolyAI was founded in 2017 by three Cambridge PhD researchers – Nikola Mrkšić, Tsung-Hsien Wen, and Pei-Hao Su – who had spent years building the academic foundations of data-driven conversational AI at the University of Cambridge's Machine Intelligence Lab. Their specific insight was that the right approach to voice AI was not to build better rules but to build better models: AI systems trained on real conversational data that could handle the full messy, interrupted, accented, digressive reality of human telephone speech. Their platform's voice agents handle calls for Marriott, Caesars Entertainment, PG&E, UniCredit, Foot Locker, FedEx, and 100+ other enterprises across 45 languages and 25+ countries, managing over 500 million calls. The company's agentic AI does the work equivalent to 1,000+ full-time employees at multiple enterprises simultaneously. Their flagship product, Agent Studio, launched in April 2025, is the platform layer that sits above the voice models. It gives enterprises full visibility into how their AI agents behave, why they respond the way they do, and how to tune their performance over time, including safety filters, tone calibration, vocabulary customisation, and real-time analytics. An enterprise can build a voice agent with a "strong Texan accent" (a real customer request, per Nikola) or fine-tune the system to handle a specific regulatory disclosure requirement in financial services. The platform is self-serve: enterprises can modify their agents without requesting changes from PolyAI, which dramatically reduces total cost of ownership and accelerates deployment. PolyAI deploys in as little as two weeks for a proof of value and two months for a full production rollout. The Series D funding, announced December 15, 2025, will be used to expand Agent Studio's capabilities and accelerate go-to-market in North America, Europe, and new verticals beyond the company's current base in hospitality, financial services, healthcare, and retail.
Why this matters
The $540B global customer service industry runs on a labour model that is structurally broken. Contact centre attrition rates average 30–45% annually in the US and regularly exceed 100% in markets with high competition for labour, meaning the average contact centre replaces its entire workforce every 12 months. Training costs, quality inconsistency, and the human cost of repetitive, high-pressure work compound into an industry that spends enormous resources simply maintaining baseline function. The Forrester Total Economic Impact study of PolyAI customers found a 391% ROI, with an average of $10.3M in savings per enterprise customer per year. But voice AI's impact on customer service is only the first-order effect. Nikola Mrkšić's more ambitious framing is the "agentic enterprise" where AI agents don't just handle inbound calls but proactively monitor the full customer and operational landscape in real time, detecting and resolving problems before human agents are even aware they exist. PG&E uses PolyAI (Peggy) to handle power outage calls at scale, including during peak emergency events when call volumes spike by orders of magnitude and human capacity is physically insufficient. The regulatory dimension is significant and underappreciated. As voice AI handles increasingly consequential interactions across payment processing, medical appointment scheduling, financial account management, the requirements for accuracy, privacy, and auditability escalate. PolyAI has built its platform for enterprise-grade compliance from the ground up, which is part of why regulated industries are adopting it faster than consumer-grade voice AI alternatives. The Citi Ventures and Point72 Ventures participation in the Series D is a direct signal that financial services, one of the most compliance-intensive and highest-call-volume industries on earth, is a primary expansion target. The competitive dynamic is an important context. Sierra Technologies (founded by ex-Salesforce CEO Bret Taylor) raised at a $10B valuation in September 2025. Decagon AI is also well-funded. PolyAI's specific differentiator is not that it is the only voice AI company but that it has been building voice-first, enterprise-grade conversational AI since 2017, has proprietary models trained on hundreds of millions of real enterprise calls, and has the deployment track record (100+ enterprise customers, 2,000+ live deployments) that new entrants will need years to replicate. The moat is the data flywheel: every call PolyAI handles makes its models more accurate, which wins more enterprise contracts, which generates more call data. That cycle has been running for eight years.
Open roles at PolyAI
6 positions we're tracking. Roles are re-checked daily and removed when filled.
ML Engineers (Speech/NLU)
First seen 4 months ago
Backend Engineers (Python/Go)
First seen 4 months ago
Solutions Architects
First seen 4 months ago
Enterprise Sales (US expansion)
First seen 4 months ago
Customer Success Managers
First seen 4 months ago
Applied Scientists
First seen 4 months ago
Know when PolyAI is hiring before anyone else
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching PolyAI
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+Hiring outlook
2000+ live deployments, 45 languages, Fortune 500 clients
Working at PolyAI
Founded in , PolyAI is Enterprise Voice Agents. They're now people. Working at a AI company at this stage means opportunity to shape your role based on the company stage.
The majority of roles are in London + SF.
How to actually get hired at PolyAI
Why applying the normal way doesn't work
There's an ATS, but at PolyAI, referrals get priority. Your cold application competes with sourced candidates and internal recommendations.
Who to contact at PolyAI
No verified contact route for PolyAI yet. Based on size () and sector (AI), the hiring manager is likely the decision-maker, but treat this as a model.
What PolyAI screens for
The Cambridge academic pipeline: PolyAI was spun out of Cambridge's Machine Intelligence Lab and all three founders did their PhDs there. The research group, led by Professor Steve Young's successors, is still the primary talent pipeline. If you are a current Cambridge PhD student or recent graduate in dialogue systems, NLU, or speech processing, you are the single most warm candidate this company sees. Email the research team directly, not through the jobs page. The Zendesk Ventures signal: Zendesk invested in this round as a strategic partner. This means PolyAI is building deep integrations with Zendesk's enterprise customer service platform. If you have Zendesk platform experience, either as an admin, developer, or solutions architect, that is a warm angle into a commercial partnership or solutions engineering role that most applicants won't have thought about. The published PolyAI research report: Nikola references PolyAI's proprietary research on voice channel usage in customer service in multiple media appearances. Finding this report (it's on the PolyAI website and has been cited in press coverage), reading it thoroughly, and referencing a specific data point in your outreach immediately signals that you've done more work than 95% of applicants. Most candidates will read the press release. Read the research. The employee to find: CTO Tsung-Hsien Wen is the most technically rigorous of the three co-founders and leads the engineering and research hiring. His academic papers are on Google Scholar. Reading his most recent published work and referencing a specific finding, not just 'I read your work' but a substantive observation, is the warmest possible cold outreach to the technical team. For commercial roles, look for PolyAI's VP of Sales or Head of Enterprise (the person filling this role is the hiring manager for AE roles and is less approached than the founders). The attrition argument: PolyAI's primary commercial argument to enterprise customers is that they eliminate the 30–45% annual contact centre attrition cycle. If you have ever managed a contact centre, worked in workforce management, or analysed attrition data in a customer service context, you have a direct experiential understanding of the problem PolyAI solves, and that lived experience is worth more in a sales or solutions role than any technical credential. Lead with that story.
A tailored CV beats a generic one. Use PolyAI's job description language to clear filters.
Don't make these mistakes
Treating voice as 'just another AI modality'. PolyAI has been doing production voice for 7 years. They've seen every failure mode. Show you understand why voice is hard.
Mistakes that kill PolyAI applications
Generic CVs stand out at a -person company — and not in a good way. Fastest path to rejection.
Don't open with what you want. Open with what PolyAI is dealing with right now — The $540B global customer service industry runs on a labour model that is structurally broken, and what you'd do about it.
The wait-and-hope strategy fails. Follow up on day five — response rates roughly double.
Applying to PolyAI? Get the contact, not the form.
The PolyAI interview process
4 stages · 14 days typical · take-home: yes · modelled from similar companies
We don't yet have verified candidate reports for PolyAI. What follows is the typical process for a -person AI company — treat it as a model, not confirmed detail.
Interview stages
Recruiter Screen
Phone or video · 30 min
Basic qualification and logistics
Recruiter or HR
Hiring Manager Interview
Video call · 45 min
Role fit and experience deep-dive
Hiring manager
Technical/Functional Round
Video call · 60 min
Skills assessment and problem-solving
Team members
Final Round
In-person or video · 60 min
Culture fit and cross-functional alignment
Senior leadership
PolyAI take-home assignment
PolyAI includes a take-home exercise in their interview process. For AI roles, this typically involves a practical problem that takes 2-4 hours. Focus on clean, working code over premature optimization. They're evaluating how you think and communicate, not just the solution.
PolyAI interview timeline
Timeline: ~ days. That's about average than the AI median (14 days at people).
Interviewed at PolyAI?
Tell us how it went — stages, questions, timeline. Takes 90 seconds and it's how this page stays accurate for the next person.
Submit your PolyAI interview experience →PolyAI jobs, frequently asked questions
How many jobs does PolyAI have open?
6 open roles at PolyAI, last checked February 2025.
Does PolyAI hire remotely?
No remote roles right now — all positions are in London + SF.
What roles is PolyAI hiring for?
PolyAI is hiring across Engineering, Other, Sales, Customer, Data. The most recent opening is ML Engineers (Speech/NLU).
How do I apply for a job at PolyAI?
Apply via the links above. For tips, read our guide on how to get hired at PolyAI.
Does PolyAI respond to cold emails?
Response rate data for PolyAI not yet confirmed.
Who is the hiring manager at PolyAI?
At this size, hiring is usually run by the hiring manager.
How competitive is it to get hired at PolyAI?
Typical applicant count for AI roles ( people): 100-250 in two weeks. Apply fast.
How many rounds is the PolyAI interview?
4 stages: Recruiter Screen, Hiring Manager Interview, Technical/Functional Round, Final Round.
Is the PolyAI interview hard?
Focus on technical depth and system design, not puzzles. Technical Interview is reportedly the most challenging round.
Does PolyAI give a take-home task?
Yes, PolyAI includes a take-home assignment.
How long does PolyAI take to get back to you?
Around 14 days across the full process.
What should I prepare for the PolyAI interview?
Focus on technical depth and system design. At people, they're testing whether you can operate without process, not whether you memorised algorithms.
Where is PolyAI based?
PolyAI is headquartered in London + SF.
Get PolyAI roles before they're posted
A role stays uncontested for about four days. Here's the window — and where we put you in it.
From $9/month, cancel any time.
Watching PolyAI
0 applicantsRole spotted & verified
1You get the alert
1You've applied
~8Hits the job boards
250+