From sci-fi to infrastructure, by design
Co-founder and CEO Nikola Mrkšić likes to joke that he has been “selling sci-fi” since his teens, first in Belgrade cafés, now in boardrooms. The ambition stuck. PolyAI builds voice agents that sound so natural callers sometimes apologise mid-call to “the person” on the other end.
That didn’t happen by accident. After Apple acquired VocalIQ to create Siri, Nikola and co-founders Tsung-Hsien Wen and Pei-Hao Su rebuilt the stack for enterprise AI voice from the ground up. Focusing on how people actually feel when they speak to a machine. The result is brand-tuned voices that match accent, vocabulary, and tone; one early demo even opened with a Texan “Howdy.”
Then the market caught up. After ChatGPT launched, inbound interest quadrupled, and pilots turned into production rollouts across hospitality, utilities, and more. “We went from being a weird company to one with a target on its back.” Nikola says in his interview for our In Focus interview series.
Why this round matters
Enterprises want automation that callers trust. Agent Studio combines lifelike dialogue with deep integration into existing systems, so voice becomes a first-class channel, not an add-on. The impact shows up for users and operators: happier callers, lighter queues, and outcomes such as 391 % ROI and $10.3m average savings per enterprise customer.
This latest funding will accelerate faster, brand-true deployments at scale. It will also deepen governance, safety, and analytics, and put richer tools in the hands of operations teams, not only ML engineers.
The bigger signal for European AI
This step up is part of a wider story: Europe can set the pace in enterprise AI, especially where trust, regulation, and design matter. Voice remains the front door for service; getting it right moves metrics PolyAI’s customers care about—containment, CSAT, and cost—without alienating users.
“In our early days, I visited a contact centre with 1,000 people,” says Nikola Mrkšić. “Today, we have customers where our AI performs the equivalent work every day. Within five years, 90% of contact-centre work will be automated. The remaining human roles will be more creative and strategic.”
Policy and investment in unison
Jensen Huang announced NVIDIA’s investment in PolyAI’s latest round on stage in London alongside UK Prime Minister, Keir Starmer. It’s a clear sign that enterprise voice now sits near the centre of the AI stack. It’s no longer left on hold.
Investment from the British Business Bank in this latest round also dials in vested interest for UK taxpayers, backing sovereign AI technology to scale at home.
What customers will feel next
More human conversations as brand-tuned voices become the default. Roll-outs measured in weeks as Agent Studio expands. Voice that matches the brand from accent to tone, because design drives trust as much as latency and accuracy. “People talk about latency and accuracy. We focus on design—how you make users feel. That’s what wins.” says Nikola Mrkšić.
The line clicks again. No hold music. Just a friendly voice.
