ElevenLabs CTO explores scaling AI agents for enterprise workflows

Deploying agents across the company

ElevenLabs is actively integrating AI agents into its core business operations to streamline various departments. The company has deployed an inbound sales development representative (SDR) agent that has managed more than 3,000 customer leads this year. This agent operates without human intervention to qualify potential clients in 32 different languages.

Beyond sales, the startup uses agents to assist with human resources and internal communications. A recruiting agent is currently responsible for conducting first-round interviews with new job applicants. Additionally, various agents are integrated into the company's internal Slack to provide immediate answers to employee questions regarding internal processes.

The company also utilizes AI as a training tool for its workforce. One specific agent serves as a coach to help new sales team members practice their skills. By allowing employees to role-play customer interactions with an AI, the company aims to accelerate the onboarding process for its sales staff.

The shift toward agentic AI value

Alex Holt, the chief technology officer at ElevenLabs, views agentic AI as a primary driver of value for technology investments. According to research from the consulting firm BCG, the share of AI value derived from agentic systems grew from 17% in 2025 to 22% this year. The report projects that this figure could rise to 39% by the year 2030.

Holt emphasizes that while these efficiency gains may seem advanced now, he expects even greater improvements in the coming months. He notes that most businesses are currently underprepared to reorganize their workflows to support multi-agent processes. These systems allow one agent to autonomously pass work to another agent to complete a complex chain of tasks.

At ElevenLabs, the goal remains tied to traditional business metrics rather than just adopting new tech for its own sake. Sales teams are expected to use these tools to generate revenue faster, while customer service teams monitor satisfaction scores closely. Holt describes the agents and technology as tools that business units should leverage to meet these specific objectives.

How the agents function and scale

The company is moving toward multi-agent systems to handle more complex work handoffs and shared context. For example, their sales lead agent uses voice AI to determine a buyer's level of seriousness before looping in a human representative. This allows the company to automate the most repetitive and mechanical parts of the sales cycle.

One customer recently completed a full sale through an interaction involving two different agents. The first agent fielded the initial inbound request, while a second account executive agent handled the legal documents and contracts. Holt explains that this allows employees to dedicate more of their time to complex requests or high-value tasks like upselling.

To ensure these systems work correctly, ElevenLabs has placed an engineer on every team across the business. These engineers work in small pods with colleagues to solve specific problems, such as reducing the time it takes to move candidates through the hiring pipeline. This structure ensures that the technology is integrated directly into the daily workflows of the staff.

Governance and error management

Using autonomous agents introduces risks, such as the possibility of the AI acting in unintended ways. ElevenLabs has addressed this by creating a human-led monitoring and observation layer to govern how its agents behave. This oversight is designed to catch mistakes before they impact the broader business or the customer experience.

Despite these safeguards, errors can still occur, such as a customer service agent providing an incorrect refund. When these instances happen, the company creates a test case to simulate the exact chain of thought and reasoning the agent used. They then update the prompts and guardrails to prevent the same mistake from happening again in the future.

Holt suggests that tracking every decision an agent makes provides a unique advantage over human-only processes. If a human makes a mistake, it might go unnoticed, but the data from an agentic mistake can be analyzed to optimize the system. He expects the next few years to see a major trend in people looking to optimize these specific processes through data collection.

Future outlook for the startup

The rapid adoption of these technologies has contributed to significant growth for ElevenLabs. After Holt joined the company in 2023, the startup announced it had doubled its valuation to $22 billion in 2026. The company is now looking toward the future, with CEO Mati Staniszewski indicating a potential initial public offering in 2028.

The company's strategy involves balancing the speed of AI deployment with the need for long-term results. Holt acknowledges that some metrics, such as whether a recruiting agent is finding the best possible hires, can take a long time to measure. He relies on business owners to identify the leading indicators that will prove the success of these initiatives.

As the technology matures, the focus will likely shift toward more complex multi-agent interactions. ElevenLabs is already developing these systems to handle more intricate handoffs between different AI personas. This evolution aims to automate entire chains of service for customers while keeping humans in the loop for the most sensitive decisions.