Saraiva AI: WhatsApp Voice to Text Automation

Saraiva AI is an experimental automation project that turns WhatsApp voice messages into intelligent text summaries using AI + no-code orchestration.

Built with n8n, Groq (Whisper Large v3), and Google Gemini, the workflow receives audio messages through the WhatsApp API (Evolution), validates permissions, and routes each message through a contextual processing pipeline.

The system automatically:

Detects whether the message comes from a private or group chat.

Applies different AI treatments depending on the duration and type of audio.

Transcribes the message using Whisper Large v3 hosted on Groq for ultra-fast inference.

Processes the text through a custom framework (OGRT – Optimization, Hook, Summary, Transcription).

Returns a structured, concise response directly to the WhatsApp user.

Short audios (≤40s) are summarized by an assistant model (“Pangeia”), producing quick actionable insights. Long audios (>40s) are processed by the “Ultron” model using the OGRT method to generate deeper analysis and formatted transcription.

Security and governance controls include:

Group permission validation and sender identification filters.

Base64 verification and timeout control for large files.

Complete LGPD-compliant design and encryption across the workflow.

The system was designed to save time, improve accessibility, and demonstrate how AI can transform everyday communication into structured, useful information — all built entirely with no-code tools.

🔗 Live Demo:

n8n (workflow orchestration)

Groq (Whisper Large v3 for transcription)

Google Gemini (NLP processing)

WhatsApp API (Evolution)

Framework OGRT (Optimization, Hook, Summary, Transcription)