Production-Ready Gmail to Jobber Request Automation (AI & GraphQL)

Overview:
We require a production-grade automation service that monitors an incoming Gmail inbox, reliably extracts structured work-order data using an LLM, and creates fully populated Jobber Requests via Jobber's GraphQL API.

Key Functional Requirements:
1. Ingestion:
- Real-time email capture using Gmail API via Google Cloud Pub/Sub webhooks (no inefficient polling).
- Robust handling of forwarded/nested email threads to extract the original requester's details, not the forwarder.

2. AI Extraction & Validation:
- LLM pipeline (OpenAI/Anthropic) using strict JSON schemas (Structured Outputs / Tool Calling).
- Extracted fields: Client Name, Phone, Email, Property Address, Job Title, Work Overview, and Job Notes.

3. Media & Snapshot Engine:
- Generate a clean PDF snapshot of the original raw email (HTML-to-PDF via Puppeteer or Playwright).
- Extract and forward incoming media/documents (JPEGs, PNGs, PDFs).

4. Jobber Sync:
- Query existing clients via GraphQL (`clients` query); match by email or phone.
- If not found, execute `clientCreate` mutation.
- Create a Jobber Request (`requestCreate` mutation) with all extracted fields, attaching the raw email PDF and media files via Jobber's file upload flow.

5. Reliability & Security:
- Deduplication / Idempotency check via a lightweight database (Redis, SQLite, or PostgreSQL) to prevent duplicate requests on webhook retries.
- Comprehensive error handling, structured logging, and fallback alerts (e.g., webhook to Discord/Slack or admin email on unparseable threads).

Deliverables:
- Clean, documented codebase (Node.js or Python).
- Setup documentation covering OAuth 2.0 configuration, Pub/Sub provisioning, and environment variables.
- Dockerfile / deployment configuration for standard cloud hosting (e.g., Railway, Render, or VPS).

Bot Filter:
Start your proposal with the keyword "PIPELINE". Bids lacking this keyword or displaying generic AI-generated filler will be archived immediately.