Automating GetMyBoat Inquiries: Building an Intelligent Inquiry-to-Booking Pipeline for JADA

What Was Done

We assessed the gaps in JADA's existing booking automation stack and designed a comprehensive inquiry handling system to bridge the manual work that Carole had been managing. The existing infrastructure—platform_inbox_scraper.py, CalendarSync.gs, and Google Calendar as the source of truth—handles confirmed bookings well. What was missing: a structured pipeline for the inquiry → response → provisional hold → payment → confirmation lifecycle.

Rather than implement everything at once, we've documented a phased automation approach that extends the current architecture without requiring a complete rewrite. The key insight: GetMyBoat inquiries are already landing in Gmail; we just need to parse them, extract structured data, and trigger downstream actions the same way confirmed bookings do.

Technical Details: The Inquiry Handler Pattern

Current Flow (What Exists):

• GetMyBoat confirmed booking → Gmail notification
• platform_inbox_scraper.py detects email, extracts booking ID, dates, customer
• Data pushed to Google Calendar iCal feed
• CalendarSync.gs (15-min poll) syncs to Calendar and triggers crew/cleaner SMS dispatch

Missing Flow (What We're Adding):

• GetMyBoat inquiry (not yet a booking) → Gmail notification
• No structured extraction or response trigger
• No provisional calendar hold while negotiating
• No auto-response or SLA tracking

Proposed Enhancement:

Extend platform_inbox_scraper.py with a new InquiryHandler class that:

1. Detects inquiry patterns in email subject/body (e.g., "GetMyBoat Inquiry" vs. "Booking Confirmed")
2. Extracts: customer name, requested dates, vessel preference, party size, special requests
3. Creates a provisional calendar entry (marked as "Inquiry – Hold") with 2-hour TTL
4. Stores inquiry metadata in a DynamoDB table (jada-getmyboat-inquiries) as source of truth
5. Sends a templated auto-response via Gmail (with operator review before sending in MVP)
6. Alerts operator via SMS if no response within 4 hours (configurable SLA)

Infrastructure: DynamoDB + Lambda + EventBridge

Rather than add complexity, we lean on AWS services already in the JADA stack:

DynamoDB Table: jada-getmyboat-inquiries

Partition Key: inquiry_id (GetMyBoat inquiry ID, e.g., "GMB-12345678")
Sort Key: created_timestamp (ISO 8601)
Attributes:
  - customer_name (string)
  - customer_email (string)
  - requested_dates_start (ISO 8601)
  - requested_dates_end (ISO 8601)
  - vessel_preference (string, nullable)
  - party_size (number)
  - special_requests (string)
  - inquiry_status (enum: "new", "responded", "provisionally_booked", "declined")
  - response_timestamp (ISO 8601, nullable)
  - calendar_event_id (Google Calendar event ID, nullable)
  - calendar_hold_expires (ISO 8601)
  - gmb_link (string, GetMyBoat direct message URL)
  - operator_assigned (string, nullable)

Lambda Function: ProcessGetMyBoatInquiry

Triggered by SNS event (from platform_inbox_scraper.py` when it detects an inquiry email):

Runtime: Python 3.11
Environment variables:
  DYNAMODB_TABLE=jada-getmyboat-inquiries
  GOOGLE_CALENDAR_ID=jada-charter@googlegroups.com
  SQS_QUEUE_URL=https://sqs.us-west-2.amazonaws.com/.../jada-responses

Handler: lambda_function.lambda_handler

Key steps:
  1. Parse SNS payload (email metadata from scraper)
  2. Call GetMyBoat API (or email regex) to extract inquiry fields
  3. Insert record into DynamoDB
  4. Create provisional Google Calendar event (title: "INQUIRY: {customer_name} – {dates}")
  5. Store calendar_event_id in DynamoDB
  6. Send SNS notification to operator (email + SMS)
  7. Return success/failure

Cron: rate(30 minutes)
Target: Lambda function ProcessInquirySLA
Logic:
  - Query DynamoDB for inquiries with status="new" AND created_timestamp < now-4h
  - For each stale inquiry, send operator SMS alert
  - Update inquiry status to "sla_violated" for reporting