ElevenLabs, a global artificial intelligence audio company, has unveiled a feature that lets users build and modify AI agents through voice or text conversation.
The company said on the 20th that it has introduced an alpha version of a new feature called ElevenAgents Architect to ElevenAgents, its enterprise voice and chat agent platform. The tool configures or changes an agent's workflows, knowledge base, prompts and voice settings in response to natural-language commands.
Architect analyzes thousands of agent conversation logs to identify problems that arose in existing workflows and proposes improvements. Suggested changes can be validated through simulation. Settings changed in the dashboard are saved as drafts and applied after user review, while changes made through external AI tools may be applied directly, depending on each tool's approval settings.
The feature also works with outside AI tools including ChatGPT, Claude Code, Cursor and Grok. Users can build and manage ElevenAgents from the AI services they already use.
ElevenLabs plans to further strengthen Architect's ability to go beyond responding to user requests and identify improvement opportunities on its own while proposing draft changes along with validation results. The company said the move is intended to reduce the work companies spend finding and fixing agent problems themselves, and to let them apply improvements continuously based on operational performance.