You write the script in Claude. Claude voices it. You get an MP3. No switching apps. No copying and pasting. One conversation.
That workflow became possible this week. ElevenLabs: the AI audio platform behind some of the most realistic voice cloning and text-to-speech technology available launched a hosted MCP server on August 17, 2026.
It plugs ElevenLabs' entire audio and creative suite directly into Claude, with one-click setup and no technical configuration required.
If you've ever produced audio content, podcasts, explainer videos, customer-facing voiceovers, or social media clips, this changes how you'll build that content going forward.
🔧 Tool of the Week: ElevenLabs Hosted MCP
MCP stands for Model Context Protocol: a standard that lets AI tools connect to each other so they can share capabilities inside a single conversation.
In layman’s terms: it means Claude can now use ElevenLabs' tools directly, without you copying output from one app and pasting it into another.
The previous version of the ElevenLabs MCP required local installation, Python setup, and an API key: technical enough that most non-developers never touched it.
The hosted MCP launched August 17 removes all of that.
You sign in with your ElevenLabs account: one click, done, revokable anytime, and the connection is live. No terminal, no API keys, no configuration files.
Here's what Claude can now do with ElevenLabs connected:
Text-to-speech across 50+ models and 3,000+ voices. Tell Claude to convert any text into audio using a specific voice from ElevenLabs' library. "Read this product description in a warm, conversational female voice" and it generates the audio file directly in the conversation.
Voice cloning. If you've cloned your own voice in ElevenLabs, Claude can generate audio in that voice. "Write an intro for this week's newsletter and read it in my voice" becomes a single instruction that produces both the script and the voiceover.
Speech-to-text transcription. Upload an audio file and Claude transcribes it with speaker identification for multi-speaker recordings. Meeting notes, podcast transcripts, customer call summaries.
Sound effect generation. Describe a sound and ElevenLabs generates it. Relevant for video content creators and anyone producing media.
Voice and chat agent management. If you run a customer-facing voice agent or chatbot through ElevenLabs, you can now review performance, update configurations, and manage it through Claude, without opening the ElevenLabs dashboard.
Video and image generation via Flows API. As of this week, ElevenLabs also added asynchronous APIs for video and image generation alongside speech, meaning the MCP now covers voice, music, video, and image production, all accessible from one Claude conversation.
The verdict: For any small business owner producing audio or video content regularly, this is the most meaningful workflow upgrade since the AI writing tools themselves. The friction of switching between tools; write here, voice it there, edit somewhere else, collapses into a single conversation. That's not incremental. It's a different way of working.
🧪 Real Business Example
A small podcast production company in Nashville: two people, six client shows per week was using a three-app workflow: Claude for show notes and scripts, ElevenLabs for ad reads and intro voiceovers, Adobe Premiere for final assembly.
The handoff between Claude and ElevenLabs alone: copy script, open browser tab, paste, configure voice, generate, download, re-upload was adding 15–20 minutes per episode and creating friction.
After connecting the ElevenLabs hosted MCP to Claude Desktop, the workflow collapsed. They write the script in Claude, ask it to generate the voiceover in the client's cloned voice, and receive the audio file in the same conversation. The 15–20 minute handoff is now under two minutes and the chances of error have eroded.
On six shows a week, that's roughly 1.5 hours saved per week just from eliminating the switching cost between two tools they were already using.
📋 Step-by-Step: Connect ElevenLabs to Claude in 5 Minutes
This is the fastest setup of any tool we've covered in this newsletter.
Make sure you have Claude Desktop installed — download at claude.ai/download if you haven't already. The hosted MCP works in Claude Desktop; the web version at claude.ai also supports it through the connectors menu.
Go to the Claude connectors directory — inside Claude Desktop, open Settings → Connectors (or look for the connectors icon in the sidebar). Search for "ElevenLabs."
Click Connect and sign in with your ElevenLabs account — one-click OAuth. No API key entry, no configuration. If you don't have an ElevenLabs account, create one free at elevenlabs.io (10,000 characters per month on the free tier).
Test the connection — open a new Claude conversation and type: "Use ElevenLabs to generate a short audio clip of the following text in a professional male voice: [any short sentence]." If the connection is working, Claude will generate and return the audio file.
Clone your voice (optional but recommended) — go to elevenlabs.io, navigate to Voice Lab, and record 1–3 minutes of yourself speaking. ElevenLabs creates a cloned voice. From that point, any Claude prompt can generate audio in your actual voice.
Try a full workflow — ask Claude to write a 60-second intro script for your business and immediately generate it in your cloned voice. That's the workflow that previously required two separate apps.
For Cursor users — run
/mcpinside a Cursor session to complete the OAuth sign-in. Same connection, same capabilities.
❓ The Dumb Question
"What's MCP and why does it matter that ElevenLabs built one?"
MCP is a standard, like a universal plug that lets AI tools talk to each other inside a single conversation.
Before MCP, every AI tool was an island: you'd use Claude for writing, ElevenLabs for voice, Canva for visuals, and spend significant time moving files between them.
MCP lets Claude reach into ElevenLabs (or any other MCP-compatible tool) and use its capabilities directly, as if they were built in.
Why it matters that ElevenLabs built one: most audio tools haven't done this yet.
ElevenLabs moving fast here and specifically prioritising Claude as the primary client; means Claude users get voice and audio capabilities that users of other AI assistants still have to access through manual tab-switching.
It's an early-mover advantage that rewards the businesses that set it up now.
💰 What It'll Cost You
Component | Cost | Notes |
|---|---|---|
ElevenLabs Free | $0 | 10,000 characters/month — enough for testing |
ElevenLabs Starter | $5/month | 30,000 characters, voice cloning |
ElevenLabs Creator | $22/month | 100,000 characters, commercial license |
ElevenLabs Pro | $99/month | 500,000 characters, professional workflows |
Claude Pro (for Claude Desktop) | $20/month | Required to use Claude Desktop with connectors |
Claude Desktop download | Free | The app itself is free |
The realistic starting point for most small business owners: ElevenLabs Starter at $5/month + Claude Pro at $20/month = $25/month total for a fully connected AI writing and voice workflow. The free ElevenLabs tier is enough to test whether this changes your workflow before spending anything.
One usage note: ElevenLabs bills per character generated, not per audio minute. A 500-word script is roughly 2,500–3,000 characters. On the Starter plan (30,000 characters/month), that's approximately 10–12 full voiceover scripts per month. Scale up to Creator if you're producing more than that.
⚡ The Practical Play
This week: connect ElevenLabs to Claude Desktop using the steps above, it takes under five minutes and use it on one real piece of content you already needed to produce.
A newsletter audio version, a product explainer voiceover, a social media clip intro.
The goal isn't to evaluate whether ElevenLabs is useful in theory. It's to run one real workflow and see whether the time savings justify the $5/month Starter cost.
For most businesses producing any regular audio content, the answer after one use will be obvious.
📰 News That Matters
The ElevenLabs hosted MCP is part of a broader pattern happening rapidly across the AI tooling ecosystem: the move from "apps you switch between" to "capabilities that connect inside one conversation."
Claude's connectors directory now includes ElevenLabs, Canva, Google Drive, Gmail, Google Calendar, and dozens of others, all accessible through the same interface, all manageable without leaving the conversation.
The practical implication for small businesses is significant: the time you currently spend moving content between tools — copying, pasting, reformatting, re-uploading is systematically becoming automatable.
This isn't happening years from now. It's happening this week.
🚫 Skip This
Trying to connect ElevenLabs to ChatGPT using third-party workarounds you find on YouTube. The email announcement mentioned ChatGPT, but the native, officially supported, no-configuration path is Claude.
Third-party wrappers that claim to bridge ElevenLabs into ChatGPT require self-hosting a remote server, maintaining it when ElevenLabs updates their API, and trusting a community-built solution with your API credentials.
Use the tool that actually supports this natively. Claude Desktop, ElevenLabs hosted MCP, five minutes. That's the setup that works.
Until next issue, Kris
The Layman's AI — The only AI updates your business actually needs.
You can check ElevenLabs out here:
The best voice models, now with full orchestration. Build real-time voice and chat agents on one low-latency stack: any LLM, your tools and knowledge, testing, Guardrails, and omnichannel deployment.
