ElevenLabs Enables Agent Transfer for Smarter Conversational AI Workflows

ElevenLabs has rolled out a powerful new feature in its conversational AI platform: agent-to-agent transfer, enabling one AI agent to pass a conversation to another based on defined, real-time conditions. This marks a significant step in designing layered AI workflows, where multiple agents can collaborate to handle inquiries with greater precision and specialization.

With the new transfer_to_agent system tool, developers can configure intelligent handoffs between agents. These transfers can occur when a user's intent matches specific criteria—like asking about billing, requesting technical help, or seeking schedule details.

For example, a general-purpose Orchestrator Agent might manage initial queries, then delegate:

• Agent 1 for availability questions,

• Agent 2 for technical support, with Agent 2a focusing on hardware issues,

• Agent 3 for billing or account matters.

This nested architecture allows organizations to deploy domain-specific agents, improving user experience by matching inquiries with the right expertise.

How It Works

To enable agent transfers in ElevenLabs:

1. Add the Transfer Tool: In the Agent tab, activate the transfer_to_agent system tool and select “Transfer to AI Agent.”

• Optional Tool Description: Customize guidance for the model on when to initiate a transfer, or leave it blank to use default logic rules.

Define Transfer Rules: For each condition, specify:

1. Agent: The designated recipient agent.

• Condition: A clear, natural-language description of the trigger (e.g., “User asks about hardware issues” or "User requests technical support for product X").

The LLM uses these rules—along with the tool description—to determine when and where a conversation should be transferred.

Note: Transfers will only work if the user configuring the system has at least viewer-level permissions for the destination agents.

ElevenLabs recommends using the gpt-4o or gpt-4o-mini models for agent transfer scenarios, due to their superior performance in tool-calling and response coordination.

Developers can also implement agent transfers via the ElevenLabs API, enabling programmatic configuration when creating or updating agents—ideal for automation and scaling.

What This Means

ElevenLabs’ introduction of agent-to-agent transfer marks a significant advancement in conversational AI. While multi-agent coordination has been explored in research and experimental settings, this appears to be the first commercially available implementation that allows AI agents to seamlessly hand off conversations based on predefined conditions.

This capability enables organizations to design modular, scalable AI systems where specialized agents handle distinct tasks—such as billing inquiries, technical support, or scheduling—while maintaining a cohesive user experience. By preserving context during transfers, the system ensures continuity and relevance in interactions, enhancing user satisfaction.

The feature reflects a broader shift toward decentralized AI architectures, where collaboration among specialized agents replaces monolithic models. As AI continues to integrate into various aspects of business and daily life, such innovations are poised to redefine how automated systems interact with users and with each other.

By making AI agents collaborative, ElevenLabs isn’t just building smarter systems—it’s building systems that work smarter together.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.