ElevenLabs launches alpha of ElevenAgents Architect
ElevenAgents Architect lets teams describe requirements by voice or text to build, test and revise AI agents, with changes saved as versioned drafts.
ElevenLabs has released the Alpha version of ElevenAgents Architect, an expert feature built into its enterprise voice and chat agent platform ElevenAgents, according to datanet.co.kr. The tool lets users build, analyze, improve and verify AI agents by describing what they need in voice or text.
ElevenAgents Architect can read and act on ElevenAgents settings, including the knowledge base, prompts, workflows, voice settings, procedures, guardrails, tools and simulations. It can also examine failed tests, sudden spikes in customer inquiries and issues flagged by ElevenAgents Spotlight, then propose and simulate changes to address them.
Any agent settings the architect modifies are stored as versioned drafts, which users review before publishing. Enterprises can set their own rules and standards that the architect must follow across every change it makes.
The feature works inside the ElevenAgents platform or through external AI tools, including Claude, Claude Code, ChatGPT, Cursor and Grok. Changes made through those outside tools may be applied directly, depending on each tool's approval settings.
Quick answers
What is ElevenAgents Architect?
It is an expert feature in ElevenAgents that builds, analyzes, improves and verifies AI agents from voice or text instructions.
Which tools can be used with ElevenAgents Architect?
It runs inside ElevenAgents and can also be used through Claude, Claude Code, ChatGPT, Cursor and Grok.
Are changes saved automatically?
Settings modified by the architect are saved as versioned drafts for users to review before publishing, though changes made through external tools may apply directly depending on those tools' approval settings.