Developers

Can Claude Make Phone Calls? Add RizzDial Connector

Claude cannot dial on its own, but you can connect it to RizzDial in claude.ai or Claude Desktop as a custom connector, no terminal required.

By James Hill, Founder, RizzDial ·

Checklist card for Add RizzDial Connector to Claude: Approve from a browser session, When Claude asks to run a specific, Reserve "Allow always" for tools, If a connection prompt shows up, "List my AI agents and tell me

Claude will not pick up a phone and dial for you. What it can do, inside the Claude app on claude.ai or in Claude Desktop, is read a phone system's data and trigger actions in it, the same way it already reads a connected calendar or drive. You get there by adding RizzDial as a custom connector: open Customize > Connectors > + Add > Add custom connector, paste the MCP server URL from your RizzDial dashboard, and approve the sign in page Claude opens. No terminal, no code, no install.

Answer first: In claude.ai or Claude Desktop, go to Customize > Connectors, click "+ Add," then "Add custom connector." Name it RizzDial, paste the remote MCP server URL from your RizzDial dashboard's Connect MCP page, and click Continue. Review the authentication Claude detected (RizzDial needs sign in), leave the OAuth client on "Use Claude's published identity," and click Add. A browser tab opens; sign in to RizzDial and approve. Turn the connector on for a chat with the "+" button, then ask something only your account can answer, such as "list my AI agents."

Verified on 2026-10-12 against the Claude custom connector docs.

What changes once Claude can reach RizzDial?

Without a connector, asking Claude about your calling setup gets a generic answer about phone systems in general, because Claude has never seen your account. With RizzDial connected, the same question gets answered from your actual agents, numbers, and call history. That is the difference between Claude describing what a voice campaign usually looks like and Claude telling you your Renewals agent has eleven missed callbacks from Tuesday.

That matters most for two step requests: "check which of my AI agents has no outbound number assigned, then tell me which area codes I should search" is one tool call for the list and a second for the number search, chained without you copying anything between screens.

What do you need before you connect?

You need a RizzDial account with the MCP connector available. If you are not a customer yet, book a call to get access first. Once you are, open the Connect MCP page in your RizzDial dashboard. It shows the remote MCP server URL for your account, and on that same page is a button that opens Claude's add custom connector screen with the URL already filled in, which skips the copy and paste step entirely.

Custom connectors work in claude.ai and Claude Desktop across every plan. The free plan allows one connector at a time. Pro and Max support more than one. Team and Enterprise add an organization wide option that an owner sets up once for everyone on the plan, covered below.

How do you connect RizzDial as a custom connector on Pro and Max?

From Claude, go to Customize, then Connectors, then click "+ Add."

Customize > Connectors > + Add > Add custom connector

Click "Add custom connector," then fill in the two fields that matter:

Name: RizzDial
Remote MCP server URL: YOUR_RIZZDIAL_MCP_URL

Click Continue. Claude inspects the server and shows you the authentication it detected. RizzDial's server needs a signed in session, so pick "Sign in now" (the other options are "Sign in when needed" and "No sign in," for servers that do not need an account at all). Below that is the OAuth client choice: keep "Use Claude's published identity (Recommended)" unless you have a specific reason to register your own OAuth client or bring one you already run. Click Add.

Claude opens a browser tab to RizzDial's sign in page. Log in with the RizzDial account you want Claude acting as, and approve the request on the screen that follows. Once approved, RizzDial shows up in your connector list as connected.

How does one owner set this up for a whole Team or Enterprise?

On Team and Enterprise plans, a connector usually gets added once and shared. The organization owner or primary owner goes to Organization settings, then Connectors, then Add, and follows the same custom connector steps above. Members of that organization then open Customize > Connectors on their own account and click "Connect" next to RizzDial, which runs their own sign in and approval without anyone needing the raw MCP URL handed to them in a message.

This is the path worth using for a non technical teammate. They never see a URL, never touch a terminal, and never need to know what MCP stands for. They click Connect, sign in to RizzDial, and approve.

How do you turn the connector on inside a single chat?

A connector being added does not mean it is active in every conversation. Claude keeps that per chat, on purpose, so an old thread does not quietly keep tool access it no longer needs. Click the "+" button in the chat box, choose Connectors, and toggle RizzDial on for that conversation. Turn it off the same way when you are done with a thread that does not need account access anymore.

What should you check before approving the connection?

Read the authentication screen before you click past it. It is the step where you decide whether Claude, running from Anthropic's cloud rather than your laptop, gets to act inside your RizzDial account. A few habits make this safer in practice:

  • Approve from a browser session already signed in to the RizzDial account you mean to connect, not whatever account happens to be logged in.
  • When Claude asks to run a specific tool mid conversation, read what it says it is about to do before you click Allow.
  • Reserve "Allow always" for tools you have seen run correctly more than once. It is a standing grant, not a one time yes.
  • If a connection prompt shows up when you did not just add or reconnect anything, deny it and go find out why it appeared.

An organization owner on Team or Enterprise can also turn off specific connector tools for everyone in Organization settings, worth doing for any tool that deletes or spends money if nobody on the team needs it from chat.

How do you verify the connection works?

A connector can be listed and still be useless if the sign in expired or the approval never finished.

First, the listing. Open Customize > Connectors and confirm RizzDial shows a connected state, not a pending or error badge. On Team and Enterprise, a member checks the same screen under their own Customize > Connectors rather than the organization settings page, since that is where their personal connection status lives.

Second, ask something only your account can answer:

List my RizzDial AI agents.

If Claude names agents you actually created, the connection works end to end. A vague answer about voice agents in general, or a claim it cannot reach RizzDial, usually means the connector is registered but not authenticated, which is a more common failure than an outright error. Remove it and add it again rather than trying to patch a partly connected state.

What prompts actually put the connector to work?

These map to what a connected RizzDial account can actually do: list and create AI agents, search and assign phone numbers, pull call history and transcripts, and start or pause campaigns.

  • "List my AI agents and tell me which ones have no outbound number assigned."
  • "Which phone numbers are available in the 512 area code, and assign one to my Intake agent."
  • "Show call history for the Renewals agent from the last seven days and summarize the most common objection."
  • "Create a new outbound agent for appointment reminders. Have it confirm the date and time twice and never mention pricing."
  • "Pause the Aged Leads campaign and tell me how many contacts were left in it when you paused."
  • "Start the Q4 Follow Up campaign." (Read this one back to yourself before you send it. It is the kind of request worth confirming first.)

Ask for a count before anything that touches many records at once. "How many contacts would that affect" turns an unnoticed mistake into a number you can check. And ask where an answer came from: a connector that can name the agent and date range behind a summary is reading your account, one that cannot is filling gaps from memory.

How do you edit or remove the connector?

There is no edit screen for a connector's URL or auth settings once it is added. If your RizzDial MCP URL changes, or you want to switch which RizzDial account Claude signs in as, remove the connector from Customize > Connectors and add it again with the new details. This is also the fastest way to clear a connector that is stuck in a broken, partly authenticated state.

Why does it matter that the connector acts as you?

Read this twice before turning the connector on in a busy chat. Once approved, Claude acts with your RizzDial permissions, and anything it changes through that connection is a real change, not a draft. "Clean up the inactive campaigns" can delete campaigns.

A few practices are worth setting up on day one rather than after something goes wrong:

  1. Connect a dedicated RizzDial user for Claude, separate from your own login, so you can see exactly what it did and cut it off without touching anyone's regular access.
  2. State the confirmation rule in a project instruction: Claude should name the change and wait for a yes before deleting, buying a number, or starting a live campaign.
  3. Turn the connector off per chat when you are not using it, rather than leaving every conversation able to reach your account.
  4. Be careful combining this connector with Advanced Research. Research runs can call connector tools across a long task without pausing for approval the way a normal chat does. Disable write capable tools, or the connector entirely, before a run you plan to step away from.
  5. Revoke from the RizzDial dashboard, not just the Claude side, if a device is lost or someone leaves. Access granted there is access revoked there.

Calls placed this way are still real calls to real people. The FCC's February 2024 declaratory ruling treats AI generated voices as "artificial" voices under the TCPA, so prior express consent rules apply the same as they would to a human caller; see the ruling on fcc.gov. RizzDial honors a contact marked do not call at the contact level, and it is built with FCC and FTC compliance in mind, but none of this is legal advice, and a connector cannot make that judgment for you before it starts dialing.

If the same team also wants an agent handling texts rather than calls, Beam's iMessage API covers that side without touching the voice connector described here.

Claude app or Claude Code: which one should you use?

If you already live in a terminal, Claude Code is the faster path and gives you claude mcp add and claude mcp list instead of clicking through settings screens. The two set up the same underlying connection in different places.

Claude app connector Claude Code
Where you set it up claude.ai or Claude Desktop, Customize > Connectors Terminal, claude mcp add
Who it suits Anyone, including non technical teammates Developers already in a terminal
Where it runs Anthropic's cloud Your machine
Turning it on Per chat, with the "+" button Registered once, available every session
Org wide setup Owner adds it once on Team or Enterprise, members click Connect Each developer runs the add command themselves

For the terminal side of this, including Codex, see Add Phone Calls to Your AI Agent with MCP. For the full set of agents RizzDial connects to beyond Claude, start at Give Your AI Agent a Phone. The RizzDial MCP page is where every one of these connector URLs actually comes from.

What do people ask about connecting Claude to RizzDial?

Can Claude actually call someone on its own?

No. Claude is a chat model, not a dialer. What it can do is read and act on a phone system through a connector, which is what you get once RizzDial is connected. Claude tells RizzDial what to do; RizzDial places the call.

Do I need Claude Code or a terminal to connect RizzDial?

No. The custom connector setup lives entirely in the Claude app, in Customize > Connectors. If you already use Claude Code, the terminal path in our Claude Code and Codex guide is faster for a developer, but it is not required.

Which Claude plans support custom connectors?

Custom connectors work in claude.ai and Claude Desktop across every plan. The free plan allows one connector at a time. Pro and Max support more. Team and Enterprise add an organization wide option that an owner manages for everyone.

Can I connect from Claude Desktop instead of the browser version?

Yes. Claude Desktop uses the same Customize > Connectors screen as claude.ai. The custom connector setup is identical in both. A separate, local setup through claude_desktop_config.json exists for running your own MCP server on your machine, but that is a different mechanism from the one covered here.

Is it safe to leave the RizzDial connector on while using Advanced Research?

Be careful with it. Advanced Research can call connector tools during a long research run without pausing for your approval on each one. Turn off any write capable tool, or the connector entirely, before starting a research run you are not watching closely.


Add the connector once, approve it once, and the rest happens in plain chat. Keep anything that deletes, buys, or goes live behind a confirmation, and point the connector at a dedicated user so you always know what Claude did and what you did yourself.

Book a call to get your RizzDial account connected to Claude.


How can RizzDial help with your calling workflow?

RizzDial is the AI outbound sales workspace for teams on GoHighLevel. Power dialing, AI voice agents, SMS automation, and CRM workflows in one platform. Book a demo.