AvulAi Developer Docs | API & Automation Integration Hub

Quickstart Guide

The AvulAi engine routes unstructured incoming events (voice calls, WhatsApp chats, emails, webhooks) through our deterministic classifier into downstream systems like HubSpot, Salesforce, and PostgreSQL.

To begin sending events to AvulAi in a sandbox environment, obtain your sandbox API token from the onboarding portal and invoke the unified ingestion endpoint.

curl -X POST https://api.avulai.com/v1/ingest/events \
  -H "Authorization: Bearer av_sandbox_sample_token_xyz" \
  -H "Content-Type: application/json" \
  -d '{
    "source": "api_client",
    "event_type": "inbound_lead",
    "payload": {
      "name": "Jane Miller",
      "company": "Apex Logistics",
      "requirement": "Schedule warehouse automation discovery call"
    }
  }'
              

Authentication

All REST API requests to AvulAi must be authenticated via Bearer tokens in the HTTP Authorization header. Each client receives separated production and sandbox API keys.

Key Prefix	Environment	Permissions
`av_sandbox_...`	Sandbox / Test	Simulation only; logs to sandbox test console without triggering live CRM mutations.
`av_live_...`	Production	Live execution; triggers connected voice telephony agents, calendar reservations, and CRM writes.

// Sample Header in all requests:
Authorization: Bearer av_sandbox_sample_token_xyz
              

POST Events Ingestion API

Unified endpoint for asynchronous event processing. Handles payload validation, intent classification, and pipeline routing with sub-500ms guaranteed execution.

Endpoint: https://api.avulai.com/v1/ingest/events

POST /v1/ingest/events HTTP/1.1
Host: api.avulai.com
Authorization: Bearer av_sandbox_sample_token_xyz
Content-Type: application/json

{
  "source": "website_chat",
  "event_type": "inbound_message",
  "payload": {
    "sender_id": "+12025550198",
    "text": "Hi, we need to integrate our Zendesk ticketing with an autonomous voice agent."
  }
}
              

Successful Response (HTTP 200)

{
  "status": "success",
  "execution_id": "exec_98a72b10f82",
  "intent_detected": "sales_demo_request",
  "confidence_score": 0.994,
  "executed_actions": [
    "hubspot_lead_created",
    "sales_slack_alert_fired",
    "calendar_invite_dispatched"
  ],
  "latency_ms": 114
}
              

POST Voice Telephony & Call Agent API

Dispatch autonomous outbound voice calls or route inbound SIP / WebRTC streams directly to your configured AI voice agent.

Endpoint: https://api.avulai.com/v1/voice/calls

POST /v1/voice/calls HTTP/1.1
Host: api.avulai.com
Authorization: Bearer av_sandbox_sample_token_xyz
Content-Type: application/json

{
  "agent_id": "agent_appointment_scheduler_v2",
  "recipient_number": "+14155552671",
  "context": {
    "customer_name": "Michael Vance",
    "service_inquired": "Commercial HVAC Maintenance",
    "preferred_window": "Tomorrow afternoon"
  },
  "compliance": {
    "recording_consent_required": true,
    "disclosure_text": "This call is assisted by an AI representative for AvulAi. Audio may be recorded for quality."
  }
}
              

Response

{
  "call_sid": "CA98124097120489",
  "status": "initiated",
  "estimated_latency_ms": 280,
  "webhook_callback_url": "https://api.avulai.com/v1/voice/webhooks/status"
}
              

Deterministic Intent Routing

Unlike raw prompt-based LLM chains that suffer from unpredictable hallucinations, AvulAi utilizes deterministic decision DAGs (Directed Acyclic Graphs). If confidence scores fall below 95%, the system immediately triggers human-in-the-loop fallback.

// Routing Schema Rule Example:
{
  "intent": "emergency_cancellation",
  "threshold": 0.95,
  "action_primary": "trigger_sms_and_calendar_free",
  "action_fallback": "escalate_to_human_queue",
  "notify_channel": "#ops-urgent-alerts"
}
              

CRM & Downstream Connectors

AvulAi includes native bi-directional connectors for enterprise data stores:

Connector	Supported Operations	Sync Latency
Salesforce / HubSpot	Create/Update Lead, Log Activity, Update Deal Stage	< 200ms
Google Calendar / Outlook	Check Free/Busy slots, Reserve appointment, Issue invite	< 350ms
Twilio / Exotel Telephony	Inbound SIP routing, Live call transfer, Call recording webhook	Sub-400ms
Slack / Microsoft Teams	Instant operational alerts, Approval buttons, Ticket summaries	< 150ms

Errors & Rate Limits

AvulAi uses conventional HTTP response codes to indicate API success or failure:

Code	Meaning	Resolution
`400 Bad Request`	Payload schema validation failed	Inspect the error JSON details for missing required fields.
`401 Unauthorized`	Invalid or missing Bearer token	Verify your API token under Settings > API Keys.
`429 Rate Limit Exceeded`	Burst limit exceeded	Implement exponential backoff. Managed tiers support up to 5,000 req/sec.
`500 Internal Gateway Error`	Downstream CRM connection timeout	AvulAi automatically queues and retries with exponential backoff up to 5 times.