Product
MCP serverREST APICaller appChatGPT, Claude & Muse
Use cases
Lead follow-upAI SDR phone stepAgencies
Developers
DocumentationQuickstartSandboxllms.txt
More
PricingGuidesSign inGet API key
Use case · AI SDR

The phone step your AI SDR can't legally take.

Your agent already researches accounts, writes sequences and handles replies. When the next best step is a call, it hands off to a person and gets the result back.

Why not an AI voice?

Since February 2024 the FCC treats AI-generated voices as "artificial voice" calls under the TCPA. Calling cell phones that way needs prior express consent, and telemarketing needs prior express written consent. Statutory damages run $500–1,500 per call. A human dialing by hand isn't an artificial voice.

Where it fits

TriggerWhat the agent does
Reply says "call me"create_call_task with the thread summary as the brief
High-intent signal (pricing page, demo no-show)Task with a short deadline_minutes
Email bounced, business has a main lineTask with consent: "b2b_main_line" to reach the right person
Outcome: bookedSend the invite and prep notes from outcome.notes
Outcome: do_not_callStop all outreach to the contact

Build it

Add the MCP server to your agent runtime, or POST to the REST API from your orchestration layer. Use external_ref to tie tasks back to your CRM record, and a webhook to resume the agent's workflow when the call is done.

// on reply.intent === "call_me"
await ringahuman.create_call_task({
  phone, state, consent: "inquiry",
  goal: "Confirm their use case and book a 20-minute demo",
  brief: summarize(thread),
  external_ref: crmContactId,
  webhook_url: "https://agent.example.com/hooks/call-outcome"
})
Give your SDR agent a phone.

Five tools. Works with any MCP runtime.

MCP setup →