Scheduling a meeting with five people across two time zones should not take six emails. Yet most organizations still rely on manual back-and-forth or simple free-busy lookups that ignore context like focus time preferences, meeting fatigue, and priority.

The Problem

Calendar tools show availability but not preference. A slot might be technically free but falls during someone’s deep work block or creates a back-to-back meeting chain. The person scheduling has no way to weigh these tradeoffs without asking everyone individually.

The AI Approach

An LLM can evaluate calendar context beyond simple availability. Feed it each participant’s calendar, their stated preferences (no meetings before 10am, prefer clustered meetings), and the meeting’s priority level. The model proposes optimal slots ranked by overall group fit.

Three-Step Build

Step 1 - Calendar integration. Connect to Google Calendar or Outlook via API to pull free-busy data and existing meeting metadata for all participants.

Step 2 - Preference-aware ranking. Pass availability windows to an LLM along with participant preferences and meeting priority. The model ranks available slots considering focus time protection, time zone fairness, and meeting clustering.

Step 3 - Propose and confirm. Send the top three options to participants via Slack or email. Auto-book when consensus is reached or escalate if no common slot exists.

Where It Breaks

Calendar data is only as good as what people actually put on their calendars. Informal commitments and ad-hoc availability changes are invisible to the system. Cross-organization scheduling adds permission complexity.

The Production Path

Start with internal team scheduling where calendar access is straightforward. Add learning from accepted vs. declined proposals to improve ranking over time. Expand to cross-team and external scheduling as confidence grows.