M
MJK.Supplies
Home / Make.com / Make.com Voice Agents: Integrate AI Phone System…
Make.com

Make.com Voice Agents: Integrate AI Phone Systems Into Your Scenarios

Voice AI agents in Make.com enable automated phone calls, WhatsApp voice messages, and AI-powered call handling — all connected to your CRM and business tools. Using Make.com with Vapi, Twilio, and ElevenLabs, you can build voice automations that qualify leads, remind customers, and provide 24/7 phone support.

M
MJK Supplies · May 23, 2026 · 11 min read
ShareXinf↗
Make.com Voice Agents: Integrate AI Phone Systems Into Your Scenarios

Voice AI Stack with Make.com

Building voice agents in Make.com:

  • Vapi — Voice AI platform; handles call orchestration, STT, TTS
  • Twilio — Telephony; Make has native Twilio module
  • ElevenLabs — Natural-sounding TTS voices
  • Claude API — AI brain for voice agent reasoning
  • Make.com — Orchestrates post-call workflows (CRM updates, follow-ups)

Vapi Webhook Integration

Vapi sends webhook events that Make.com processes:

Set up in Vapi:

  • Assistant settings → Server URL → your Make.com webhook URL

Events Make.com receives:

  • call.started — new inbound call
  • call.ended — call completed (with transcript + summary)
  • tool_calls — agent wants to execute a tool (e.g., lookup customer, book meeting)
  • hang — call detected hang state

Make.com Webhook trigger:

  1. Add Webhook module to your Scenario
  2. Copy the webhook URL
  3. Paste into Vapi's server URL setting

Post-Call Workflow

The most common Vapi + Make.com integration: process call results after the call ends.

Scenario: Post-call CRM update

  1. Webhook trigger: Receive call.ended event from Vapi
  2. Filter: Only process if status = "ended" (not cancelled, failed, etc.)
  3. Tools → Parse JSON: Extract transcript, summary, call duration, caller number
  4. HTTP (Claude): Analyse call
Analyse this sales call transcript: {{transcript}} Return JSON: { "outcome": "interested|not_interested|callback_requested|wrong_number|voicemail", "key_points": ["list of important points from call"], "next_action": "specific recommended next step", "urgency": "high|medium|low", "contact_name": "if mentioned", "contact_email": "if mentioned" }
  1. Parse JSON
  2. Router: Branch by outcome

- interested → HubSpot: create deal + Slack alert to sales rep - callback_requested → HubSpot: create follow-up task + schedule callback - voicemail → Airtable: log + schedule retry in 24h - not_interested → HubSpot: update contact status

  1. HubSpot: Log call activity with transcript and AI summary

Outbound Call Campaign

Scenario: Launch outbound campaign

  1. Trigger: Airtable new row (campaign launch row added)
  2. Airtable: Get all leads in the campaign list
  3. Iterator: For each lead
  4. HTTP (Vapi API): Initiate outbound call
POST https://api.vapi.ai/call/phone Headers: Authorization: Bearer {{vapi_token}} Body: { "phoneNumberId": "{{vapi_phone_id}}", "customer": {"number": "{{lead.phone}}"}, "assistantId": "{{campaign_assistant_id}}", "assistantOverrides": { "firstMessage": "Hi {{lead.firstName}}, this is Alex calling from [Company]..." } }
  1. Wait: 5 minutes before next call (rate limiting)
  2. Airtable: Log call initiated + Vapi call ID for tracking

Tool Call Handling

When a Vapi voice agent needs real-time data (e.g., check availability, look up order status), Make.com handles the tool call:

Vapi tool call → Make.com → external system → return to Vapi

  1. Webhook: Receive tool_calls event from Vapi
  2. Router: Switch on function name

- check_availability → Calendly API: get open slots - lookup_order → Database: query order status - book_appointment → Calendly API: book + HubSpot: create deal

  1. HTTP (Vapi): Return tool result
POST https://api.vapi.ai/call/{{call_id}}/response Body: { "toolCallResults": [{ "toolCallId": "{{tool_call_id}}", "result": "{{tool_result_json}}" }] }

This happens in under 1 second — the voice agent waits for the response.

Appointment Reminder Calls

Automated voice reminders:

  1. Schedule: Hourly
  2. Airtable: Find appointments in next 24 hours with no reminder sent
  3. Iterator: For each appointment
  4. HTTP (Vapi): Initiate reminder call with custom script
  5. Vapi webhook → separate Scenario: Handle response (1 = confirm, 2 = reschedule)
  6. Airtable: Update appointment status
  7. Calendar: Confirm or cancel based on response

SMS Fallback for Missed Calls

If the voice agent can't reach someone:

  1. Vapi webhook: Call ended with outcome = voicemail or no answer
  2. Twilio: Send SMS follow-up
Hi {{name}}, we tried calling but missed you. Reply YES if you'd like us to call back, or visit {{link}} to book a time that works.
  1. Make.com (SMS reply webhook): Process reply
  2. If YES: schedule callback; if no reply in 24h: add to lower-priority list

Recommended Tools

  • Make.com — Voice workflow orchestration
  • Vapi — Voice AI platform
  • Twilio — Telephony infrastructure (native module)
  • ElevenLabs — Natural voices for Vapi agents
  • Claude API — AI for call analysis
  • HubSpot — CRM for post-call updates
#make.com#voice#twilio

Related articles

MJK Supplies · Automation Services

Want this built for you?

We design and ship custom AI agents and automation systems for teams that want results, not a backlog. Book a free 30-minute consult — no commitment, no pitch deck.