Luke Harries

Luke Harries leads growth at ElevenLabs, working across product, marketing, and developer experience to make AI-generated speech useful in real products. His career spans reinforcement-learning research, healthcare entrepreneurship, developer tools, and voice interfaces for AI assistants.

Harries studied pre-med at Cambridge and worked on reinforcement learning at Microsoft Research before co-founding Fella Health, a Y Combinator-backed healthcare startup. He subsequently became interim head of product at PostHog and joined ElevenLabs as its growth lead. His early experiments with voice-controlled assistants already combined language-model orchestration with tools that could operate beyond a chat interface.

• Voice Engine without rebuilding the agent. Harries advocates adding speech to existing assistants while preserving their orchestration, retrieval systems, evaluations, and tools. His Voice Engine demonstration integrates Scribe speech recognition, Eleven v3 speech synthesis, emotion-aware turn-taking, and semantic voice activity detection. Server and client SDKs connect the existing agent to browser interfaces, telephony, and contact-center systems; backend tools remain in place, while client-side tools can interact directly with the page.

• WhatsApp MCP server. His open-source WhatsApp integration gives AI assistants access to messaging through the Model Context Protocol. A Go bridge, Python server, and local SQLite database support searching conversations, retrieving context, and sending messages or media. Its documentation also addresses the security risks created when agents combine private information, untrusted messages, and externally consequential actions.

• Engineers should talk directly to customers. Harries argues that firsthand user research improves technical judgment and product decisions. He favors small teams with clear ownership and develops products around customer pull and research push: understanding what users need while identifying capabilities newly unlocked by stronger models.

• Growth as product infrastructure. His writing treats startup search optimization and product launches as coordinated engineering, distribution, and positioning challenges. At ElevenLabs, he has also detailed the licensing, contributor payouts, usage-based billing, and international compliance needed to turn synthetic voices into a functioning marketplace.