Developers
Every conversation, programmable.
Read any call over REST, run the whole platform from your MCP client, and send every outcome to your systems.

Front deskLinden Animal Clinic
Phone · 8:14 AM · 03:24
ExampleREST API
curl \
$API/ v1/ call_session/ retrieve/$ID \
-H "Authorization: Bearer $TOKEN" \
-H "X-Organization-Id: $ORG"200 · session
{ "session": {
"status": "ended",
"end_reason": "agent_hangup",
"variables": [{
"key": "appointment",
"value": "Thu 9:40 AM" }],
… } }MCP server
https://
youDid yesterday’s limping-pet calls get booked?
- tool
list_calls{ "searchQuery": "limping" } - tool
get_call{ "sessionId": "3f6c2a91-…" } - assistantYes. Hannah Kerr’s beagle, Biscuit, is booked for Thu 9:40 AM with Dr. Lund, and the confirmation went out by text.
Webhooks
POST hooks.lindenanimal.example/
- event
- Summary ready
- attempt
- 2 · signed
{
"phone": "+1 212 555 0118",
"owner": "Hannah Kerr",
"pet": "Biscuit, beagle",
…
"slot": "Thu 9:40 AM",
"visit_id": "LV-4471"
}204Sent
Your API
Import your API. The agent calls it mid-call.
Point Zumu at your OpenAPI spec, or describe one call by hand. Each operation becomes a tool the agent uses while the customer is still on the line.
Your spec
openapi: 3.1.0
paths:
/visits:
post:
operationId: book_visit
summary: Book a visit
requestBody:
content:
application/ json:
schema:
required: [phone, pet, reason, slot]
properties:
phone: { type: string }
pet: { type: string }
reason: { type: string }
slot: { type: string }Imported as a tool
- Hannah
Thursday morning is fine, the earlier the better.
- Tool
book_visitslot "Thu 9:00 AM" · pet "Biscuit, beagle"409 slot_taken - Agent
That time was just taken. The next opening is 9:40. Shall I book it?
- Hannah
Yes, please.
- Tool
book_visitslot "Thu 9:40 AM" · pet "Biscuit, beagle"201
BookedVisit LV-4471 · Thu 9:40 AM
Build Zumu into the systems you already run.
After your demo call, your team has working credentials and its first integration planned.
book_visit slot "Thu 9:40 AM"201Bookedvisit LV-4471
