Balderton x ElevenLabs Worldwide Hackathon

Voice is becoming one of the most important frontiers in AI

The first voice applications (like call-centre AI) have now reached production scale across industries, and early evidence shows that voice-native interfaces can deliver a magnitude of improvements in efficiency and user experience over chat-only systems. In the YC F25 batch, 22% of startups now have a voice-native UI as part of their product. And naturally, this surge in application-layer voice companies is driving dramatic innovation at the infrastructure level, enabling better performance across all variables – with things like human-like prosody, latency under 300ms, and multimodal grounding moving into the mainstream. And 2025 has only been the start.

That’s why we at Balderton are so excited about voice AI – both on the application side, voice infrastructure, and the model layer. We couldn’t be more delighted to partner with our friends at Eleven Labs and support the community of builders who push the envelope of what’s possible in voice.

Together with our friends at Eleven Labs, we just helped throw the largest-ever builder event for voice AI globally

The London hackathon was electric. Over 30 teams gathered at Balderton HQ, bringing an unmistakable hacker energy: whiteboards filled with diagrams, teams joined forces on the spot to tackle the most ambitious ideas together, and and (proverbial) GPUs hummed late into the night. London is increasingly becoming one of the world’s top cities for AI talent, and last Thursday was another reminder why.

We were honoured to have an outstanding judging panel – including Mati Staniszewski from ElevenLabs, Harry Stebbings from 20VC, Balderton’s own Laura McGinnis (our consumer AI guru), Pippa Thompson, Co-Founder Asymmetric Security, and Zafeer Ahmed, CRO at Ebury – helping spotlight the most exciting ideas.

But London was just one part of a what can only be described as a global coordination feat by the ElevenLabs team: 1300+ participants simultaneously hacking across 33 cities worldwide, with hundreds of voice projects created during the largest voice AI hackathon ever. It was a privilege to see London play its part in that global movement.

The London hack’s winners showed just how broad and impactful voice AI can be

The first-place team built a real-time gaming copilot – an in-game companion that players can speak to mid-battle, mid-mission, or mid-chaos, offering strategy, guidance, and support that adapts to what’s happening on screen. The second team created a simulation platform for employee onboarding and upskilling: a set of role-specific voice agents that replicate real-world conversations in sales, customer support, HR and more, helping employees train faster and more effectively than traditional programmes. And the third team tackled one of the most meaningful applications of all: a mental-health triage agent, built around a virtual avatar, that speaks with patients, interprets what they say – and how they say it – and provides early-stage guidance on what to do next.

If you’re a founder building anything in voice AI — from pre-seed to pre-IPO — we would love to meet you. You can reach me at [email protected].