MCP: Use Vocal Video From Claude, ChatGPT and Other AI Assistants

Connect Claude, ChatGPT or another AI assistant to Vocal Video, and you can request, review, edit and share videos by asking for it in the chat you already use. Your assistant works with the same tools as Vera, the AI assistant built into Vocal Video.

💡 Note: Connecting an AI assistant is included on Essential, Pro, Scale and Enterprise plans.

In this article:

Step 1: Copy your server URL

Connecting takes three steps and a couple of minutes. First, copy your server URL: the web address your assistant uses to reach your Vocal Video account. It's the same for every assistant.

  1. In Vocal Video, go to Settings > Connect AI Assistant (MCP).
  2. Click Copy beside MCP Server URL.

The Connect AI Assistant (MCP) settings page, with the Copy button beside the MCP Server URL marked and the Toolsets menu set to All

The Toolsets menu above the URL sets which tools your assistant can use. All is selected, and it's the right choice for most assistants. If yours limits how many tools it loads, choose Collect + Edit, which leaves out galleries and review shares, and copy the URL again.

Step 2: Add it to your assistant and sign in

Each assistant adds the connection a little differently, but they all end the same way: a Vocal Video sign-in where you choose a workspace and click Allow. Follow the steps for yours.

Claude

The connection lives on your Claude account, so it works in Claude on the web and in the Claude Desktop app.

  1. In Claude, go to Customize > Connectors, click +, then Add custom connector.
  2. Enter Vocal Video as the name and paste your server URL into Remote MCP server URL.
  3. Leave Advanced settings closed and click Add. You don't need a client ID or secret.
  4. When Claude opens a Vocal Video sign-in, log in, choose a workspace and click Allow.
  5. In a chat, click +, then Connectors, and switch on Vocal Video.

Claude's Add custom connector dialog, with Vocal Video as the name and the server URL field marked

On a Claude Team or Enterprise plan: an Owner adds the connector first under Organization settings > Connectors > Add > Custom > Web. Each member then finds it under Customize > Connectors and clicks Connect.

On Claude's free plan: you can add one custom connector.

ChatGPT

In ChatGPT, custom connections only appear once Developer mode is on. On a ChatGPT Business, Enterprise or Edu workspace, an admin needs to enable it first.

ChatGPT's Developer mode setting switched on, with its Elevated risk label and the toggle marked

  1. Go to Settings > Security and login and switch on Developer mode. ChatGPT labels it elevated risk because it lets you add connectors OpenAI hasn't reviewed. That's expected.
  2. Click Plugins in the left sidebar, then click + next to the search box.
  3. Enter Vocal Video as the name, choose Server URL and paste your server URL, then set Authentication to OAuth.
  4. Leave Advanced OAuth settings as they are. Tick I understand and want to continue and click Create.
  5. When ChatGPT opens a Vocal Video sign-in, log in, choose a workspace and click Allow.
  6. In a new chat, click + in the message box and pick Vocal Video.

ChatGPT's New Plugin dialog, with Vocal Video as the name, the server URL entered and Authentication set to OAuth, marked

ChatGPT asks you to confirm each change before it makes it.

Claude Code

Claude Code adds the connection with one command, which already includes the server URL.

  1. In a terminal, run claude mcp add --transport http vocal-video "https://vocalvideo.com/api/mcp?toolsets=all"
  2. Start a Claude Code session, type /mcp  and pick Vocal Video.
  3. When the Vocal Video sign-in opens in your browser, log in, choose a workspace and click Allow.

Cursor

  1. Before you copy your server URL in Step 1, choose Collect + Edit in the Toolsets menu. Cursor only uses the first 40 tools it's given.
  2. Add the server URL to your mcp.json  file. Our developer docs have the exact lines to add.
  3. When Cursor asks, approve the Vocal Video sign-in: log in, choose a workspace and click Allow.

Other assistants

Any assistant that supports remote MCP servers can connect with the same server URL, and most open the same Vocal Video sign-in. See Other MCP clients in our developer docs, including setup for assistants that use an API key instead of a sign-in.

Step 3: Try your first request

In a chat with Vocal Video switched on, ask something simple to check the connection, like "List my collectors." Your assistant may ask your permission the first time it uses Vocal Video. Once it answers, you're connected. See What you can ask for for what to try next.

What you can ask for

Once it's connected, ask in plain words, using your collector, video and gallery names as they appear in Vocal Video. Here's a sample of what your assistant can do.

Request videos

  • "Create a collector called Customer Stories with three questions about the results our customers have seen."
  • "Add a question asking what they'd tell a colleague about us, and make it the first question."
  • "Write a video request for the Customer Stories collector and send it to the people in this list."
  • "Pause the Hiring collector."

Review responses

  • "Summarize what people said in my Pricing collector."
  • "Which responses in my Pricing collector mention ROI?"
  • "Star the three strongest answers to the second question in the Q3 collector."

Make and edit videos

  • "Make a draft video from Dana's response using the LinkedIn Feed template."
  • "Build a highlight reel from the starred responses in the Customer Stories collector."
  • "Tighten the pauses in the Acme testimonial and remove the filler words."
  • "Fix the spelling of Acme everywhere in the captions."
  • "Make the Acme testimonial vertical for Instagram Reels, with bold word-by-word captions."
  • "Set our default template to end every video with our logo animation."

For how templates shape each video, see Video Templates: the Complete Guide.

Share videos

  • "Send the Acme draft to Sam for review."
  • "Give me the share link and embed code for the Acme testimonial."
  • "Create a gallery of our five newest customer videos and give me the embed code."
  • "Swap anything older than 90 days in our homepage gallery for newer videos."

For more on galleries, see Video Galleries: the Complete Guide.

Manage your brand and account

  • "Set our brand colors to navy #1B2A4A and coral #FF6B5B."
  • "How many AI credits do we have left?"

Combine steps in one request

Your assistant can string several steps together, so one request can do the work of many clicks:

  • "Find this quarter's responses that mention onboarding, star the best three, and make a draft video from each."
  • "Create a collector for new customers with two questions about why they chose us, then write a video request and show me the preview before you send it to the people in this spreadsheet."
  • "Build a highlight reel from this month's starred responses, tighten it, and send the draft to Sam for review."

For more prompts to copy, and recipes for agents that run on a schedule, see How to Run Your Video Testimonial Program From Claude, ChatGPT or Any AI Agent.

Anything that emails real people, like a video request or a draft shared for review, is previewed first, and your assistant asks before it sends. Videos your assistant builds start as drafts. Publishing is always your click.

Run tasks on a schedule

Some assistants can run a task on a schedule, or work through a job on their own while you're away. Connected to Vocal Video, they can keep your videos moving without you in the chat. For example:

  • "Every Monday morning, find last week's new responses, star the strongest, make a draft video from each, and send me a summary with links to the drafts."
  • "On the first of each month, swap anything older than 90 days in our homepage gallery for newer videos."
  • "Every Friday, tell me which collectors got no responses this week."

Set the task up in your assistant like any other scheduled task. Your Vocal Video connection stays signed in until you disconnect it.

Decide what it may send. In a chat, your assistant shows you a preview of any email and waits for your OK. A task running on its own has no one to ask, so say in its instructions whether to send or to leave the preview for you, for example "write the video requests and send me the previews."

Drafts still wait for you. A scheduled task can build and edit drafts, but only you can publish.

For automations that start the moment something happens, like a new response arriving or a video being published, Zapier and webhooks are often the better fit. See Integrations and Automation: the Complete Guide.

Is it secure?

  • You sign in on Vocal Video. Your assistant never sees your password.
  • Access is short-lived. Your assistant gets a temporary token that renews on its own.
  • It works as you, in one workspace. Your assistant works in the workspace you chose at sign-in, with your role's permissions. Teammates each connect with their own Vocal Video login, so there's nothing to share.
  • Nothing sends without your OK. Video requests and review shares are previewed first. For a scheduled task, you decide in its instructions (see Run tasks on a schedule).
  • You can remove access anytime. See Disconnecting your assistant below.

Does it use AI credits?

Not on its own. Your assistant's conversation runs on your assistant's plan, and most of what it does in Vocal Video is free. Two things spend credits, just as they do in the app: AI-directed templates and videos with AI effects. See AI Credits: What Uses Them and What Happens When You Run Out.

Video requests your assistant sends count toward your plan's video requests, the same as sending them from the app.

Disconnecting your assistant

  1. In Vocal Video, go to Settings > Connect AI Assistant (MCP). Each assistant you've connected to this workspace is listed with the date it connected.
  2. Click Disconnect next to the one you want to remove.

You can also remove it from your assistant: in Claude, from Customize > Connectors; in ChatGPT, from Plugins. Either way, your assistant's access stops right away, and it will need to sign in again to reconnect.

Frequently asked questions

Do I need an API key?

No. Your assistant connects with a sign-in, including for scheduled tasks. API keys are for software you build yourself, and for assistants that can't open a sign-in. They're included on Pro plans and above. See Integrations and Automation: the Complete Guide.

Can my teammates use my connection?

No. Each person connects with their own Vocal Video login.

Can reviewers connect an assistant?

Yes, to find and look through responses and videos, and to star responses. The Connect AI Assistant (MCP) settings page needs an Editor role, so connect with this server URL instead: https://vocalvideo.com/api/mcp?toolsets=all . To make changes, ask a workspace editor or admin. Changing brand settings needs an admin.

Which workspace does my assistant use?

The one you choose when you sign in.

Can my assistant publish videos?

No. It builds and edits drafts, and you publish them.

How is this different from Vera?

Vera is built into Vocal Video. Connecting your own assistant lets you do the same kind of work from Claude, ChatGPT or another assistant you already use, and lets you run it on a schedule.

For more information see:

Still need help? Contact Support at support@vocalvideo.com