How It Works

Processing Workflow

Docfire AI Engine follows a three-phase flow:

1. Ingest (Reception)

The client sends a POST /api/ingest request with:

• The template to use (e.g., translate, pdf_extraction, seo_description)
• The products (array of JSON objects) to process
• The template-specific parameters (e.g., source/target language)

An enrichment session is created with a pending status.

2. Process (Processing)

An asynchronous job:

• Builds the appropriate AI prompt based on the template
• Sends the data to the AI model (Google Gemini)
• Parses the response and saves the results to the database
• Updates the session status to completed or failed

3. Results

The client can:

• Poll the status — GET /api/sessions/{id}/status to check progress
• Retrieve results — GET /api/sessions/{id}/results to get the enriched data