Inside ElevenLabs and the Micro-Team Revolution: Why Small Teams Are Winning the AI Race
In AI, speed and depth of research decide the race. Everything else is downstream.
AI’s newest organisational shift is clear: away from layered departments, towards small, autonomous, cross-functional teams. Call them pods. Call them micro-teams. The principle is the same: five to ten people, end-to-end ownership, minimal overhead.
Few companies embody this shift more clearly than ElevenLabs.
Even after passing 250 employees, the company is organised into ~20 micro-teams. In Product, each team owns a defined product area, from the studio interface to enterprise-grade voice agents, and is designed to move with the speed of a startup.
This is not a stopgap until “real” departments appear. It is a deliberate, durable strategy.
The proof is visible. ElevenLabs has surpassed $200M ARR and expects to exceed $300M by year-end 2025, outpacing rivals with far larger teams, while consistently leading third-party benchmarks on voice AI quality.
Talent density and micro-team autonomy are the real performance levers.
ElevenLabs’ Mini Pods
ElevenLabs’ leadership is explicit about its philosophy: more people rarely fix the problem. What matters is dense talent, small teams, and the ability to move fast and iterate.
Each pod functions like a mini-startup: high autonomy, rapid loops, minimal bureaucracy. Engineers are not just coders. They clear a steep technical bar, sketch flows in Figma, draft system architecture, ship, learn, and loop.
This model also rewrites the classic Product Manager role. Specs, roadmaps, and hand-offs slow AI down. At ElevenLabs, engineers own the product end-to-end.
Why it works:
Speed and agility: Small groups communicate faster, decide quicker, and pivot without delay.
Ownership and accountability: Teams that own a product area care more, and it shows in quality and engagement.
Innovation and creativity: Autonomy plus cross-functionality creates fertile ground for new ideas.
Talent attraction: Top builders want impact. The chance to own an area end-to-end is magnetic.
The trade-off: as pods multiply, coordination and knowledge-sharing get harder. Silos creep in. Success depends on investing in connective tissue: lightweight processes, shared rituals, and strong communication norms that keep the teams aligned.
Hiring at Scale Without Lowering the Bar
Scaling at speed can mean diluting standards. That is not an option here.
Every engineer must clear the same path: tough coding bar, product challenge, flow design, architecture sketch. Even at 250+ people, founders still interview every hire. To make it workable, AI recruiting agents “AI Becky” and “AI Oscar” handle FAQs and prep, while humans focus on the highest-leverage calls.
The outcome is clear: if you get in, you know you are joining one of the most selective AI teams in the world.
Remote-First, by Design
ElevenLabs has been remote-first since day one to chase scarce researchers wherever they are. To make it work, they run on asynchronous excellence: crisp documentation, minimal meetings, communication built for speed.
The team is intentionally global, with GTM and engineering hubs in London, New York, and Warsaw, and roles listed in San Francisco and Tokyo.
Minimal, but Clear
At ElevenLabs, titles are treated as a distraction. Internally, if decisions stall everyone knows who will lead. Externally, titles appear only when they serve a purpose (for customers or cultural context).
That does not mean leadership is invisible. New joiners can step up quickly and run functions. The rule is simple: tenure does not grant authority. Contribution does.
In the early scaling phase, this stripped away status games and kept decisions close to the work. At 400, the risk is fuzziness: decisions stall, authority hides in shadows, speed suffers.
The company knows this and has built internal mechanisms: when teams cannot agree on a path forward, there is a clear decision matrix. That is not bureaucracy. It is clarity, and it is rare.
Flat here does not mean static. The promise is growth: forge your own path, progress on strengths, and learn alongside some of the best AI builders in the world.
This is organisational minimalism with one purpose: don’t let structure slow the science.
What to Emulate
First-principles thinking → solve from scratch, not default.
Mission-driven product → a voice in every language, accent, and dialect.
Tiny teams, real ownership → five to ten people, end-to-end.
Titles take a back seat → status is noise, contribution earns scope.
Engineer-led product → code, design, ship, loop.
Remote-first to chase talent → go where they are, then get out of their way.
Protect the research edge → stay months ahead, not weeks.
What to Watch Out For
Founders as bottlenecks → interviews preserve the bar until they slow the funnel.
Decision fog → flat works at 100, risks chaos at 400.
Pods drifting apart → Conway’s Law means 20 pods can become 20 products.
Culture stretch → global growth needs rituals so the culture travels beyond the founders.
Point of View
In AI, speed and research depth are the only real moats. Organisation design either protects that edge or erodes it.
ElevenLabs is betting on micro-teams, minimal titles, and a fierce hiring bar to scale speed and depth. OpenAI is making a similar bet with its pod model. This pattern is shaping the next generation of AI companies.
This is not just an AI story. It is a story about the next organisational revolution.
So here’s the founder’s test: if you woke up with 400 people tomorrow, would your company still be this fast?
About
I’m Federica De Cillis, a leadership and organisational architect and founder of Arc Studio.
My work focuses on how decisions inside companies — about structure, incentives, control, and leadership — compound over time into outcomes.
I write Margin Call for founders, operators, and investors who are curious about how companies actually work beneath the surface: how optionality is created or lost, how judgment is tested under uncertainty, and why smart teams often end up where they do.
Disclaimer
This analysis is based entirely on publicly available sources: interviews, podcasts, articles, and content shared by ElevenLabs. I haven’t worked with the company directly; this is my independent POV on what they’ve made public.