BUILT FOR PROPTECH TEAMS

Embed AI Voice Calling Into Your Platform. Go Live in a Day.

Trusted by PropTech teams, brokers, CRM platforms, and AI-first real estate companies.

Launch AI calling directly inside your platform

<3s
API to phone ringing
180ms
Median speech latency
99.9%
Uptime SLA
50
Free sandbox calls
POST /v1/calls/trigger
Authorization: Bearer zp_live_••••••••

{
  "lead_phone": "+"91-9876543210",
  "lead_name": "Amit Sharma",
  "property_id": "sector-65-3bhk-launch",
  "language": "hinglish",
  "qualification_script": "residential_buyer_standard_v2",
  "calendar_endpoint": "https://your-crm.com/api/availability",
  "webhook_url": "https://your-platform.com/zappio/events"
}
200 OK — Response
{
  "call_id": "zp_call_2826374abc",
  "status": "initiated",
  "estimated_ring_time_seconds": 28,
  "lead_id": "lead_98234",
  "project_id": "sector-65-3bhk-launch"
}

That is the complete integration for an outbound AI qualification call. Your first live call fires within 2 minutes of receiving your API key.

Why Build This Yourself?

Building voice AI for Indian real estate from scratch costs 12 months and a specialist team.

We know because we built it. Here is what you are looking at if you attempt to build this capability internally.

Speech Recognition for Indian Accents and Hinglish

Standard ASR models — Google Speech, Whisper, Azure — fail meaningfully on Hinglish, regional Indian accents, and real estate domain vocabulary. "Possession", "carpet area", "OC received", "lock-in period" require domain-trained models. Training your own accurate model requires tens of thousands of hours of labeled Indian real estate call audio. We spent two years building this.

Sub-200ms Latency in Production

Real-time voice conversation requires the complete pipeline — speaker detection, transcription, LLM inference, and TTS synthesis — to complete in under 200 milliseconds per exchange. At 300ms the conversation feels robotic. At 500ms, users hang up. Achieving reliable sub-200ms latency on Indian network conditions requires significant infrastructure investment and continuous tuning. Our median latency is 180ms.

Concurrency at Launch Scale

A new project launch in Gurgaon or Mumbai can generate 400–600 simultaneous lead form submissions within the first 2 hours. Your infrastructure needs to initiate 400 concurrent calls, each with its own speech pipeline, without quality degradation. This is a non-trivial distributed systems problem. Zappio's infrastructure is built for this. Yours does not need to be.

Your product roadmap has 40 things on it. Building production-grade AI voice infrastructure from scratch is a 12-month distraction from all of them. Zappio solves this in a day.

What the API Gives You

Every capability you need. Clean, documented, and tested.

01 — Outbound Call Trigger

Fire an AI qualification call within seconds of any lead inquiry

Specify the language preference, qualification script, calendar API endpoint, and webhook URL. One API call. Everything else is handled.

KEY PARAMETERS
lead_phone
lead_name
property_id
language (hi / en / hinglish)
qualification_script
webhook_url
calendar_endpoint
POST /v1/calls/trigger
POST /v1/calls/trigger
# Language: hi | en | hinglish
# Script: your custom or standard

{
  "lead_phone": "+91-9XXXXXXXXX",
  "language": "hinglish",
  "qualification_script": "res_buyer_v2",
  "webhook_url": "https://your.app/events"
}

# Response in < 3 seconds:
{ "status": "initiated", "call_id": "zp_call_..." }
INBOUND CALL FLOW
1

Buyer clicks "Call Builder" on your portal

2

Call routes through your Zappio number

3

AI picks up instantly — no IVR, no wait time

4

AI identifies caller if known lead, begins context-aware conversation

5

Full transcript + score webhook fires on completion

02 — Inbound Call Handling

Every incoming call handled instantly with full context

Route incoming calls to your Zappio AI agent with a simple number forward. The AI picks up instantly, identifies the caller if they are a known lead, and begins a context-aware qualification or follow-up conversation. No wait time. No IVR menu.

Use case

Your real estate portal has a "Call Builder" button. Route all clicks through a Zappio number. Every call is handled, recorded, and transcribed automatically — zero manual effort.

03 — Real-Time Webhook Events

Every call state fires a webhook within 500ms

Build your own UI, trigger downstream workflows, update your CRM, or send notifications based on what happens in every call — without polling. 3 retries with exponential backoff. HMAC-SHA256 signature verification.

Webhook events — full list
call.initiated— AI call started dialing
call.answered— Lead picked up
call.qualification_started— AI began qualifying
qualification.question_asked— AI asked a question
qualification.answer_captured— Lead answered
lead.score_updated— Score recalculated
appointment.offered— AI offered site visit slot
appointment.booked— Slot confirmed by lead
call.escalation_requested— Lead asked for human
call.completed— Call ended, transcript ready
call.no_answer— Retry logic triggered
GET /v1/calls/{call_id}/transcript
{
  "call_id": "zp_call_2826374abc",
  "duration_seconds": 187,
  "lead_score": {
    "overall": 88,
    "budget": 92,
    "timeline": 85,
    "intent": 90
  },
  "transcript": [
    {"speaker": "ai", "text": "Namaste! Main Zappio...", "ts_ms": 0},
    {"speaker": "lead", "text": "Haan, budget 85 lakh...", "ts_ms": 14200}
  ]
}
04 — Transcript and Score API

Full structured transcript and per-dimension lead score after every call

Speaker-labeled transcript with millisecond timestamps and confidence scores for ambiguous responses. Lead score object with individual scores for budget, timeline, intent, and decision authority — plus a composite. Use the data to populate your CRM, display summaries in your UI, or run your own downstream scoring models on top of our raw data.

05 — Custom Qualification Scripts

Define fully custom question flows in JSON with conditional branching

Support for linear scripts, conditional branching, language-specific response variations, and multi-level routing logic. Up to 20 questions with nested conditional trees. Define a script per property segment — residential buyers, commercial lessees, NRI investors, builder plot purchasers — and reference the correct script ID in each API call.

Custom script definition — JSON
{
  "script_id": "residential_buyer_ncr_v3",
  "language": "hinglish",
  "questions": [
    {
      "id": "q1_budget",
      "ask": "Aapka approximate budget kya hai?",
      "capture": "budget_range",
      "branch": {
        "above_1cr": "q2_timeline",
        "below_1cr": "q2_timeline_alt",
        "unclear": "q1_budget_clarify"
      }
    }
  ]
}
CALENDAR INTEGRATION FLOW
API CALL INCLUDES
calendar_endpoint: "https://your-crm.com/api/availability"
DURING CALL
Zappio fetches real available slots live
AI OFFERS LEAD
"Saturday 11 AM or Sunday 3 PM — which works?"
LEAD CONFIRMS
"Saturday 11 AM"
ZAPPIO FIRES
appointment.booked webhook with confirmed time
YOUR SYSTEM
Updates calendar via your own existing logic
06 — Calendar Integration

Pass your availability API — Zappio books directly during the call

Pass your platform's availability API endpoint in the call trigger request. Zappio fetches real available slots during the conversation and offers them to the lead directly. When the lead confirms, Zappio fires a booking webhook. Your calendar is updated through your own system. No separate calendar integration to build or maintain.

Under the Hood

The infrastructure behind every call.

SPECIFICATION
VALUE
API call to call initiated
Under 3 seconds
Speech response latency
Under 200ms median
Languages supported
Hindi, English, Hinglish (production) + 8 regional (beta)
Concurrent calls per account
Up to 500 (enterprise: custom)
Monthly uptime SLA
99.9%
Call recording format
MP3 and WebM, accessible via signed URL
Transcript format
JSON with speaker labels and millisecond timestamps
Lead score output
JSON object — per-dimension scores and composite
Webhook delivery
Within 500ms, 3 retries with exponential backoff
Sandbox environment
50 free test calls — full production parity
Authentication
Bearer token (API key)
Standard rate limit
100 calls/min, 2,000 calls/hour
Data residency
AWS ap-south-1 (Mumbai) — mandatory, not optional
SDK availability
Node.js, Python, PHP + Postman collection
Webhook signature verification
HMAC-SHA256
Build with Zappio

What does your product need AI calling for?

Real Estate CRM

Add AI Calling to Your Real Estate CRM as a Core Feature

Your CRM manages lead data, pipeline stages, and agent assignments. What it cannot do is call every lead within 30 seconds of form submission and run a 15-point qualification conversation in Hindi and English. Zappio adds that through 4 API endpoints and a webhook listener.

When a new lead enters your CRM, your system triggers a Zappio call. When the call completes, Zappio fires the transcript and score back to your webhook. Your CRM displays the qualification summary directly in the lead record. You ship a capability that would have taken your team 12 months internally — in a week.

Integration points
# On new lead creation:
POST /v1/calls/trigger

# Pull summary into lead record:
GET /v1/calls/{call_id}/transcript

# Trigger lead record update:
webhook: call.completed

# Create calendar entry:
webhook: appointment.booked
Real Estate Portal

Give Builders the First-Response Advantage as a Premium Feature

Property portals compete on lead volume. The next differentiator is lead speed. When a buyer submits a property inquiry on your portal, Zappio calls them within 28 seconds — before any agent on any other platform has seen the notification.

Offer this as a premium listing tier. Builders who enable AI instant response on your portal pay a higher subscription. They get higher lead contact rates. You get higher retention and competitive differentiation.

Integration points
# On lead form submission:
POST /v1/calls/trigger

# Show quality score to builder:
webhook: lead.score_updated

# Show confirmed visits in analytics:
webhook: appointment.booked
Builder Internal Platform

AI Calling Infrastructure for an Enterprise Builder's Internal Stack

Large builders running 8–15 active projects simultaneously often have internal tech teams building custom platforms. These teams need AI calling infrastructure that handles high concurrency, supports custom qualification logic per project, and integrates without compromising data security.

Zappio's API gives your internal team the calling infrastructure without the ASR training, TTS tuning, latency engineering, and Indian accent model work. You focus on business logic. We handle the voice layer.

Integration approach
# Custom script per project:
Script configuration API

# Segmented by project:
Project-level API key management

# Route to your internal event bus:
Webhook → Kafka / SQS / Pub/Sub

# Data residency — verifiable:
AWS ap-south-1 (Mumbai) — default
From Zero to First Call

Your first live AI call in under 15 minutes.

01
10 minutes

Create your sandbox account

Access the sandbox through the button on this page. You receive your sandbox API key immediately. No credit card. No approval process. No sales call required before you can test. The sandbox environment is full production parity — same AI models, same latency targets, same webhook behavior. The only difference is that calls are flagged as sandbox and do not count toward your live plan quota. You get 50 free test calls to explore the complete integration.

02
3 minutes

Trigger your first test call

Copy the code example from the quickstart in your preferred language — Node.js, Python, or PHP. Drop in your sandbox API key. Set the lead_phone to your own mobile number. Hit send. Your phone will ring within 28 seconds and an AI will qualify you as a real estate lead. You will hear exactly what your users' leads will hear.

03
2 minutes

Set up your webhook listener and verify events

Point a webhook endpoint — you can use a free tool like Webhook.site or ngrok for testing — to your Zappio webhook URL configuration. Make the same test call again. Watch the events arrive: call.initiated, call.answered, qualification events, call.completed. Review the transcript and score object in the final event payload. That is the complete data your system will receive for every production call.

After setup: Full API reference documentation, complete Postman collection, code samples in Node.js, Python, and PHP, and integration guides for LeadSquared, HubSpot, and Salesforce are available from the moment you create your account.

FAQ

Technical questions we get most often.

Yes. Every Zappio developer account includes 50 free test calls in the sandbox environment. The sandbox runs on identical infrastructure to production — same AI models, same speech pipeline, same latency targets, same webhook event structure. You can build and test your complete integration before writing a single line of production code or committing to a paid plan.

Under 3 seconds from API call to the lead's phone beginning to ring in production. This includes queue processing, number provisioning, and call initiation. Speech response latency — the time between the lead finishing a sentence and the AI beginning its response — is under 200ms median, which is within the threshold for natural conversational feel.

In production with high accuracy: Hindi, English, and Hinglish. In beta with good accuracy and active improvement: Marathi, Tamil, Telugu, Bengali, Gujarati, Punjabi, Kannada, and Malayalam. If your platform serves a specific regional market, contact us to discuss accuracy benchmarks for that language before going to production.

You can define fully custom qualification scripts in JSON with conditional branching, language-specific response variants, and multi-level routing logic. Scripts can have up to 20 questions with nested conditional trees. You can create and manage multiple scripts — one per property segment or use case — and reference the appropriate script ID in each API call. You can update scripts through the API without any downtime.

All call audio, transcripts, and lead data are stored exclusively in AWS ap-south-1 (Mumbai region). Indian data residency is not an optional add-on or a higher-tier feature — it is the default configuration for all Zappio accounts. This is verifiable through our infrastructure documentation and available for review by your legal or compliance team at any time.

The AI is designed to handle ambiguity gracefully. For unclear responses, it asks for clarification in natural language — the way a human caller would. If it encounters a conversation state it cannot handle — a complaint, a language it was not configured for, or an escalation request — it transfers the call to your configured fallback number, logs the point of failure in the transcript, and fires a call.escalation_requested webhook event. We review all escalation events weekly to improve the AI's handling of edge cases.

Get API Access

Your first AI call should take 15 minutes, not 12 months.

Access the sandbox. Trigger a test call to your own number. Read the transcript that comes back. That is the complete integration experience — and it takes 15 minutes from creating your account.

Read Full API Documentation
India data residency included
99.9% uptime SLA
Dedicated technical onboarding
Postman collection included
50 free test calls — no credit card