AI Integration for FareHarbor and Google Calendar

The integration surface spans three key data objects: FareHarbor bookings (with their associated activity, start_time, and guide fields), Google Calendar events, and the operational staff/guide records. AI agents monitor the FareHarbor API for new, updated, or cancelled bookings. For each change, the system must decide whether to create, modify, or delete a corresponding Google Calendar event, mapping booking details to event fields like summary, location, attendees (the guide's email), and description. The primary challenge isn't the initial sync, but handling the inevitable conflicts—like double-booking a guide or last-minute schedule changes—where AI provides deterministic logic beyond simple if/then rules.

An AI layer adds intelligence to this sync in three high-value workflows:

• Conflict Resolution & Proactive Alerts: When a new booking attempts to assign a guide already scheduled in Google Calendar, the AI evaluates the conflict's severity (e.g., overlapping vs. adjacent bookings), checks guide certifications and preferences, and can either automatically reassign the booking to an available guide or escalate a Slack alert to a manager with resolution options.
• Context-Aware Reminders: Instead of generic calendar alerts, the AI crafts and sends personalized SMS or email reminders to guides 24 hours before a tour. It pulls in relevant context from the FareHarbor booking—like customer_count, special_requests, or weather_forecast—and inserts it into the reminder (e.g., "Heads up: 12 guests, 2 with mobility needs, sunny forecast").
• Batch Schedule Optimization: For operators managing multiple guides across dozens of daily tours, an AI agent can run nightly, analyzing the next day's Google Calendar events against FareHarbor bookings. It identifies potential inefficiencies (e.g., a guide with excessive travel time between locations) and suggests optimized reassignments, presenting the revised schedule for human approval before making any calendar updates.

Rolling this out requires a phased approach. Start with a one-way, read-only sync from FareHarbor to Google Calendar to establish the data pipeline and audit logs. Then, introduce AI for conflict detection and notification only, keeping humans in the approval loop. Finally, enable automated resolution for low-risk conflicts (like reassigning a backup guide with identical certifications). Governance is critical: all AI-driven actions should be logged to a separate audit table with the booking_id, action_taken, confidence_score, and human_override_flag. This ensures operators can review and revert decisions, building trust in the system. For teams already using tools like n8n or Make, this AI logic can be implemented as a custom decision node within existing automation workflows, connecting via FareHarbor's webhooks and the Google Calendar API.

The core integration is built on a middleware layer that listens to FareHarbor's webhook events (e.g., booking.created, booking.updated, booking.cancelled) and Google Calendar's push notifications. This layer, typically a cloud function or microservice, maintains the sync state and executes the AI logic. Key data objects flow bidirectionally:

• From FareHarbor to Calendar: Booking ID, product name, start/end times, guest count, guide assignment, and custom fields.
• From Calendar to FareHarbor: Event status changes (e.g., declined, tentative), updated titles, and modified time slots from guide calendars.

The AI layer sits within this middleware to manage three critical workflows:

1. Conflict Resolution & Smart Placement: When a new booking is created, the AI evaluates the assigned guide's Google Calendar for existing appointments, travel time, and working hour policies. If a conflict is detected, it can automatically propose an alternative time to the FareHarbor operator or reassign the booking to another available guide using logic from the Bokun guide coordination system.
2. Dynamic Reminder Generation: Instead of static reminders, the AI crafts context-aware notifications. It analyzes the booking's product type (e.g., "Sunset Kayak Tour"), location, and forecasted weather to generate and send personalized SMS or email reminders via Twilio or Mailchimp integrations, including what to bring and where to meet.
3. Exception Handling & State Sync: The AI monitors for mismatches—like a guide declining a calendar event after assignment—and triggers automated remediation workflows in FareHarbor, such as updating the booking status and notifying the operations manager via a Slack integration.

For rollout, we recommend a phased approach: start with a one-way sync (FareHarbor to Calendar) for a pilot guide team, then introduce the AI conflict resolver, and finally enable the full two-way sync with exception handling. Governance is critical; all AI-driven changes should be logged to an audit trail in a system like Datadog or Splunk, and major actions (like automatic reassignments) should be configured to require human approval during the initial stabilization period. This architecture ensures the integration acts as a reliable operational assistant, not an unpredictable automation.

A production-grade integration connects to two critical systems: the FareHarbor API for booking data and the Google Calendar API for calendar management. Security is enforced via OAuth 2.0 for Google and API keys/service accounts for FareHarbor, with all credentials managed in a secrets vault. The sync logic operates as a middleware service, listening for FareHarbor webhooks (booking.created, booking.updated) and processing them through a queue to handle retries and avoid data loss. Each calendar event includes metadata linking it back to the specific FareHarbor booking_id, product_id, and guide_id for traceability.

The core AI function is conflict resolution. When a new booking is received, the service first fetches the guide's Google Calendar to check for existing events. For simple conflicts (e.g., double-booked time slots), the AI can propose alternative times based on the guide's other availability or similar product schedules. For complex multi-guide or resource conflicts, the AI flags the booking for human review in a dashboard, providing context and recommendations. The system also powers smart reminders, using the booking's product details and customer notes to generate personalized pre-trip messages sent via SMS or email 24 hours before the event.

Rollout follows a phased approach: Phase 1 involves a read-only sync for a pilot group of guides, logging proposed changes without writing to calendars. Phase 2 enables write access for non-conflicting bookings with a manual approval step in Slack or Microsoft Teams. Phase 3 introduces full automation for a defined rule set (e.g., auto-accept bookings >48 hours out with no conflicts), while maintaining an audit log of all actions and a simple override command (/cancel-sync) for guides. Governance includes weekly reviews of conflict resolution accuracy and reminder engagement metrics, with prompts and logic tuned based on real operator feedback.