Company Spotlight: ElevenLabs Builds the Voice AI Stack

ElevenLabs is building the voice layer for software, but the company is no longer just a text-to-speech vendor. Its platform now spans conversational agents for enterprises, creative tools for media production, and APIs that let developers embed audio models into products.

Co-founders Mati Staniszewski and Piotr Dąbkowski started the company in 2022 after growing up in Poland and noticing how poor dubbing could flatten the emotion of films. Four years later, ElevenLabs says more than 800 employees across more than 50 countries are working on a much broader problem: making interaction with technology feel natural across languages, channels, and use cases.

The company matters now because voice AI is moving from demos into operational systems. ElevenLabs says its agent platform handles more than 15M conversations a week, while enterprise customers account for 55% of revenue. The next contest is not simply who can generate the most realistic voice. It is who can turn speech, reasoning, integrations, monitoring, and safety controls into reliable infrastructure.

From Voice Generation to an Interaction Stack

ElevenLabs organizes its product portfolio around three platforms. ElevenCreative gives creators and marketers tools for speech, dubbing, music, images, and video across more than 70 languages. ElevenAPI exposes its audio models to developers. ElevenAgents lets organizations build voice and chat agents connected to knowledge, tools, and business workflows.

That architecture expands the addressable market. A creator might begin with narration or dubbing. A developer might add speech to an application. An enterprise team might deploy agents for support, scheduling, sales, payments, or public services. The same research base can serve each path, while the application layer captures more of the workflow around the model.

It also changes what ElevenLabs has to be good at. Model quality remains central, but production customers evaluate latency, reliability, security, observability, integrations, data handling, and whether a system knows when to hand work to a person. That is a larger operating surface than voice generation alone.

Why ElevenLabs Matters Right Now

The commercial pace is difficult to ignore, even with the proper caveat that the metrics are company-reported rather than audited public disclosures. ElevenLabs says ElevenAgents annual recurring revenue has grown more than threefold since February 2026 and weekly conversations have exceeded 15M. In a September interview with TechCrunch, Staniszewski said the company had reached $600M in ARR, with more than 55% coming from enterprise customers.

The capital market has followed that trajectory. ElevenLabs raised a $500M Series D at an $11B valuation in February 2026, then completed a $300M employee tender offer at a $22B valuation in September. The tender was secondary liquidity for employees and existing holders, not automatically $300M of fresh operating capital, but it still showed investor demand for ownership at twice the February valuation.

The signal is broader than valuation. Enterprise adoption suggests conversational systems are becoming a new interface for routine business processes. When a voice agent can identify a customer, retrieve account context, take an authorized action, and document the result, the product competes with parts of contact-center software, workflow automation, and application infrastructure at once.

Leadership Built Around Research and Deployment

Staniszewski serves as CEO after working at Palantir Technologies. Dąbkowski is CTO and previously worked at Google. The pairing matters because ElevenLabs is trying to connect frontier-model research with enterprise deployment rather than treating them as separate businesses.

Its founding story is product-specific: two friends who disliked the quality of dubbed films built tools intended to preserve the texture of human speech across languages. The current company is more ambitious, but that original problem still appears in its product logic. Voice is not just an output format. Tone, pacing, timing, identity, and emotional context all affect whether a synthetic interaction feels useful or alienating.

The company’s published operating values emphasize first-principles thinking, excellence, and ownership of outcomes. Those values read less like branding when matched against the technical range of the platform. A team building agents for healthcare, government, financial services, and telecommunications has to own consequences far beyond whether generated audio sounds convincing.

Safety Is Part of the Product Surface

Voice AI carries obvious misuse and trust risks. ElevenLabs describes a layered approach that includes pre-deployment testing, runtime guardrails, production monitoring, content provenance, abuse detection, data controls, and independent validation. Its agent tooling includes redaction, configurable retention, data residency, and private deployment options for sensitive workloads.

The company also argues that organizations should disclose when a customer is speaking with an automated agent. That question will become more important as speech systems improve. A more natural interface increases utility, but it can also make the boundary between human and machine harder to perceive.

For operators, safety is therefore not a policy page sitting beside the product. It is part of the enterprise buying decision. Systems that speak for a company need controls for identity, authorization, escalation, and review, especially when conversations lead to financial, healthcare, or government actions.

Hiring Momentum as a Market Signal

ElevenLabs’ careers page listed 173 open positions across 53 locations when DevCuration reviewed it on October 1, 2026. Roles spanned core engineering, research, AI safety, security, enterprise deployment, creative products, growth, operations, and revenue.

The mix is more informative than the raw count. Hiring across model development and forward-deployed customer work suggests the company is scaling both the technology and the implementation layer around it. Openings for regional deployment and go-to-market teams also point to a market that must be adapted to different languages, regulations, industries, and communications norms.

Job listings are not a promise of future headcount, and counts can change quickly. They are still a useful snapshot of organizational priorities. ElevenLabs appears to be investing in the hard middle between a strong model and a durable production system: integrations, safeguards, rollout support, and customer-specific execution.

The Bigger Industry Shift

ElevenLabs sits inside a market where boundaries are blurring. Foundation-model companies are moving into applications, application companies are training models, and customers may become competitors as they build more of their own stack. That pressure means voice quality alone is unlikely to remain a permanent moat.

The stronger position is a compounding system: better models attract developers and customers; usage produces workflow knowledge; deployment experience improves reliability and controls; and a broader platform gives users more reasons to stay. ElevenLabs is assembling that system across creative production, APIs, and enterprise agents.

The company still has to prove that rapid growth can become durable economics while maintaining trust in high-stakes deployments. Its private valuation does not answer questions about margins, retention, or customer concentration. What it does show is how quickly voice has moved from a media-generation feature toward a primary interface for software.

For founders, operators, and investors, ElevenLabs is worth watching because it is testing a consequential thesis: the next major software interface may not be another screen. It may be a conversation, and the winning platform will need to make that conversation useful, accountable, and operational at global scale.