When a learner enrols in your Edesy LMS course, an AI voice agent calls within minutes, welcomes them by name, checks they can log in and shares the first class time. A 2-minute call costs about ₹12 at ₹6 a minute, plus telephony. The LMS API doesn't share phone numbers yet, so the number comes from your records.
Works with: Edesy LMS via webhooks in n8n or Make, with the phone number from your enquiry form, sheet or CRM · via API from your own server · Native LMS to Voice: planned
Note A native welcome-call option in Edesy LMS (and learner phone numbers in its API) is planned. Request early access.
How it works#
enrolment.created webhook ──> n8n: verify x-edesy-signature
{ learnerId, courseId } │ GET /api/v1/learners -> name, email
│ GET /api/v1/courses -> course name
│ phone: your sheet / CRM by email
▼
POST /api/v1/calls ──> welcome call ──> call.analyzed ──> sheet / counsellor
Before you start#
- An Edesy LMS site at lms.edesy.in.
- An LMS API key with
learners:read(keys start withedk_live_), from Site → Settings → Developer. - The learner's phone somewhere you can look up by email: the Google Sheet behind your admission form, your CRM, or an Edesy Forms enquiry form.
- A voice agent, its ID, and a voice agent API key (
vp_live_...).
Set it up (n8n)#
- In n8n, add a Webhook node, method POST, Respond: Immediately. Copy the Production URL.
- In Edesy LMS, open your site's Settings → Developer. In the Webhooks panel, click Add endpoint, paste the URL as the Endpoint URL, tick
enrolment.createdunder Which events?, and save. Copy the signing secret. - Verify each delivery in a Code node. The header is
x-edesy-signature: t=<unix seconds>,v1=<hex>, an HMAC-SHA256 of<t>.<raw body>; reject anything older than 5 minutes. - Look up the learner and course. The API has list endpoints only, so page through and match on ID:
GET https://lms.edesy.in/api/v1/learners # data[]: id, name, email, enrolments; nextCursor
GET https://lms.edesy.in/api/v1/courses # data[]: id, name, slug, published
Authorization: Bearer edk_live_xxxxxxxx
Match data.learnerId to a learner's id and data.courseId to a course's id. The course title is in name.
5. Find the phone: a Google Sheets lookup on the learner's email in your admissions sheet, or a CRM search. Stop if there is no number or no call consent.
6. Add an HTTP Request node:
POST https://voice-agent.edesy.in/api/v1/calls
Authorization: Bearer vp_live_xxxxxxxx
Idempotency-Key: enrol-{{ $json.learnerId }}-{{ $json.courseId }}
Content-Type: application/json
{
"agentId": 1042,
"phoneNumber": "{{ $json.phone }}",
"variables": {
"customer_name": "{{ $json.name }}",
"course_name": "{{ $json.courseName }}",
"first_class": "Monday 7 pm",
"support_whatsapp": "+91 98XXX XXXXX"
},
"retry": { "enabled": true, "maxAttempts": 2, "delayMinutes": 120, "dispositions": ["NO_ANSWER", "BUSY"] }
}
- Activate the workflow.
Simpler path if you control the sign-up form: if learners enrol through your own form or checkout that already has the phone, start the call from that event instead (see Form submission to AI call) and skip the LMS lookup.
Data map#
| Direction | Field | Meaning |
|---|---|---|
| From LMS | event, createdAt, data.learnerId, data.courseId, data.reason |
Who enrolled in what, and why (for example a purchase or an admin grant) |
| From LMS API | learner name, email; course name |
No phone in the LMS API today |
| From your records | phone |
Looked up by email |
| Into the call | customer_name, course_name, first_class, support_whatsapp |
Used in the prompt |
Back in call.analyzed |
disposition |
SUCCESS (welcomed, can log in), CALLBACK_REQUESTED, WRONG_NUMBER, DO_NOT_CALL |
| Back | extracted_data |
For example can_login (Boolean), preferred_language, question_for_team |
| Back | call_summary |
For the course coordinator |
Sample agent prompt#
You are Kavya from Vidya Academy, Lucknow. {{customer_name}} has just enrolled in
"{{course_name}}". This is a short welcome call, not a sales call.
Speak in Hindi by default; switch to English if they reply in English.
1. Congratulate them by name on joining {{course_name}}.
2. Ask if they have logged in to the course app. If not, tell them to use the email they
enrolled with and the "Forgot password" link, and note can_login = false.
3. Tell them the first live class is {{first_class}} and the recording will be in the app.
4. Ask if they have any question; note it for the team. Do not answer fee or refund questions
yourself; say the coordinator will reply on WhatsApp at {{support_whatsapp}}.
Keep it under two minutes. Set SUCCESS when done, CALLBACK_REQUESTED if they are busy,
DO_NOT_CALL if they don't want calls.
Test it#
- Enrol a test learner whose email is in your admissions sheet with your own phone number.
- Check the n8n run found the learner, course and phone.
- Answer the call, say you can't log in, and confirm
extracted_data.can_loginisfalsein thecall.analyzeddelivery.
Consent, DND and calling hours#
- Ask for consent to be called on the enrolment or admission form ("I agree to receive calls about my course"), and only call learners who gave it.
- Edesy does not scrub against the NCPR (DND) registry. A welcome call is about the learner's own enrolment; don't upsell on it without consent.
- For school-age learners, call the parent's number given at admission.
- Set Settings → Call Settings → Calling Window, for example 10:00–20:00 IST for students.
- Add anyone who opts out to Settings → Blocked Numbers.
What it can't do yet#
- LMS webhooks carry IDs only, and the LMS API returns no learner phone, so the number must come from another system.
- The learner and course APIs are list-only; large schools page through many records per lookup.
- The call result isn't written back to the learner's LMS profile.
FAQ#
Which languages can the welcome call use? Hindi, English and 13 more Indian languages, including Tamil, Telugu, Bengali, Marathi and Kannada, plus Hinglish. See languages.
How soon after enrolment does it call?
Within a minute or two of enrolment.created, inside your calling window.
Can it also call learners who stop studying? Not from an LMS event yet: there is no "stalled learner" webhook. You can run a scheduled campaign from a CSV of inactive learners. See campaigns.
Can it call after course completion for feedback?
Yes, with the same flow on enrolment.completed and a short survey prompt.
What does it cost per learner? About ₹12 for a 2-minute call at ₹6 a minute, plus telephony, on pay as you go.
Related#
- Blog: AI calls for course enquiries and coaching
- Form submission to AI call
- Appointment reminder call
- Call results to Google Sheets
- Course builder · Education use case · AI voice agent
Start free · Book a demo · Don't see your app? Request it