ElevenLabs CLI v1 Puts Entire Voice Platform in Terminal

ElevenLabs CLI v1 arrived Aug 24, and it does something simple but ambitious: it puts the entire ElevenLabs API behind a single terminal interface.

The release leads with agents. Every endpoint in the OpenAPI spec becomes a subcommand that returns structured JSON, and you get a --dry-run preview before anything actually changes.

What the terminal interface actually exposes

Text to speech, speech to text, dubbing, sound effects, music, and ElevenAgents are all callable as elevenlabs service verb. Think agents list --format json or music compose --prompt.

For coding agents, the CLI adds --schema, which returns a machine-readable contract for any command. It also handles errors in two ways: parseable JSON for agents and human-readable hints for TTY users.

Why agents-as-code changes the deployment model

The workflow is straightforward. You run elevenlabs agents pull to materialize every workspace agent as local config files, edit them on disk, then push with --dry-run for a diff preview and push again to apply.

Branches keep dev and prod separate, so a test phone number and a cheaper LLM can run in development while production sticks with the real number and a frontier model. agents init scaffolds new projects from templates like customer support.

This is built for scale, not a handful of agents sitting in a dashboard. When you manage fleets across hundreds of orgs or bulk migrate from another provider, version control becomes the control plane.

How this shifts pressure across voice and video tooling

Voice is no longer just a dashboard or an API call. It is now versioned code that lives where coding agents already work.

StartupHub.ai data positions ElevenLabs as the infrastructure leader in voice, and the CLI widens that gap by moving from audio generation into developer workflow and system of record. That matters against creative tooling vendors that still sell seats and studios.

Runway, the video generation platform StartupHub.ai tracks, raised $315 million at a $5.3 billion valuation led by General Atlantic after raising $308 million at just over $3 billion earlier in 2025. Synthesia, the enterprise avatar video player, raised $200 million at a $4 billion valuation, doubling its $2.1 billion mark from its $180 million round.

HeyGen raised $60 million at a $500 million valuation, bringing total funding to $74 million. Descript was reported to be raising at around $550 million, more than double its $260 million valuation in 2021.

ElevenLabs itself raised $500 million at $11 billion led by Sequoia Capital, more than tripling its $3.3 billion valuation from the $180 million Series C in January. That funding scale, paired with the CLI, signals a deliberate move up the stack from voice model to agent platform.

The batteries-included piece is skills. elevenlabs generate-skills installs workflow guides that teach a coding agent how to build an agent, generate speech, or transcribe a call without guessing the sequence.

The limitation is real, though. Agents-as-code assumes git, file access, and terminal control, which helps engineering-led teams but adds friction for no-code operators who live in the dashboard.

If agents stay in the terminal, voice infrastructure that fits that habit wins the next integration cycle.