1. Introduction & Location
Welcome to the Aromatic Intelligence Privacy Policy. We are an AI voice automation company physically headquartered and operating in India.
This document outlines exactly how our backend captures, stores, and processes your data, including phone numbers, call audio, and appointment details, as well as the third-party infrastructure partners we rely on to deliver our human-like AI agents.
2. Data We Collect & How It Is Used
When you interact with our AI agents, our backend systematically saves the call data. The primary data we collect includes:
- Call Conversations & Audio: The complete audio and transcribed text of conversations with our AI agents.
- Appointment Details: Names, phone numbers, and the specific reason for booking an appointment.
Training & Quality Assurance: We actively use and review the saved backend data to train our AI models, improve conversation latency, and continuously enhance the user experience and internal workflows.
3. Third-Party Subprocessors
To operate our real-time voice AI, we share necessary segments of your data (like audio streams or conversation text) with specialized third-party infrastructure providers:
- ChatGPT (OpenAI) & Groq: We use ChatGPT for core conversational logic, and Groq's high-speed LLM inference engine to ensure the agent talks with minimal latency.
- Deepgram & Google: We utilize Deepgram for hyper-fast Speech-to-Text (STT) transcription of your voice, and Google's Text-to-Speech (TTS) to generate the AI's spoken responses.
- Vobiz: Used for secure phone number provisioning, routing, and overall call handling.
- Activepieces: Used to manage and automate our complex backend data workflows.
- Google Meet / Calendar: When our AI successfully books a new appointment, the appointment data (including the caller's phone number, name, and reason for the call) is explicitly saved to Google Meet and Google Calendar.
4. Data Security & Retention
While data is processed by the third parties listed above for operational functionality, we ensure that all data transmission occurs over secure, encrypted channels. Any data used to improve our models and workflows is processed securely and is stripped of personally identifiable information where applicable.