Skip to main content

Overview

The update_preferences tool updates a user’s scheduling preferences, which directly influence how the AI ranks and selects meeting times. This is powerful for:
  • Customizing work hours
  • Setting energy patterns (morning person, etc.)
  • Configuring buffer time between meetings
  • Defining focus blocks for deep work
Updated preferences take effect immediately for all future scheduling.

Tool Schema

Usage in Claude Desktop

User: “I want to keep Fridays meeting-free for deep work” Claude will:

Response Format

Success Response

Example Workflows

Workflow 1: Setting Work Hours

Workflow 2: Energy Pattern Optimization

Workflow 3: Focus Blocks

Workflow 4: Buffer Time Management

Workflow 5: Natural Language Processing

Preference Fields Explained

work_hours

Defines when you’re available each day:
  • enabled: Whether you work this day
  • start: Work start time (24-hour format)
  • end: Work end time (24-hour format)
Meetings will only be scheduled during these hours.

buffer_minutes

Gap between consecutive meetings:
  • 0: Back-to-back meetings allowed
  • 15: 15-minute buffer (recommended)
  • 30: 30-minute buffer (for busy executives)

availability_window_days

How far ahead the AI searches:
  • 7: Next week only (urgent scheduling)
  • 14: Two weeks (default, recommended)
  • 30: Next month (flexible scheduling)
  • 90: Three months (long-term planning)

max_meetings_per_day

Daily meeting limit to prevent burnout:
  • 1-3: Light meeting load
  • 4-6: Moderate meeting load (default: 5)
  • 7-10: Heavy meeting load
AI will avoid scheduling beyond this limit.

energy_pattern

When you’re most productive:
  • flexible: No preference (default)
  • morning_person: Best before noon
  • afternoon_person: Peak 12-5pm
  • evening_person: Best after 5pm
Affects time slot scoring significantly (±20% score adjustment).

scheduling_context

Free-form text that the AI processes to understand your preferences: Good examples:
  • “I prefer to keep Fridays light for planning and reflection”
  • “Morning meetings work best for strategic discussions. Save afternoons for execution.”
  • “Avoid back-to-back meetings - I need transition time”
  • “No meetings before 10am - I have family commitments”
The AI uses Claude LLM to extract insights and generate SmartWeights.

focus_blocks

Protected time for deep work:
AI will strongly avoid scheduling during these blocks.

How Preferences Affect Scoring

Preferences directly influence the 0.0-1.0 quality score for each time slot:

Work Hours (Required)

  • Inside work hours: No penalty
  • Outside work hours: Slot excluded entirely

Buffer Minutes

  • Buffer met: +10% to score
  • No buffer (back-to-back): -15% to score

Daily Meeting Limit

  • Under limit: No penalty
  • At limit: -20% to score
  • Over limit: Slot excluded

Energy Pattern

  • Morning person + morning slot: +20% to score
  • Morning person + evening slot: -20% to score
  • (Similar for other patterns)

Focus Blocks

  • Overlaps focus block: -50% to score (strong avoidance)

Scheduling Context

  • Processed by AI into SmartWeights
  • Can adjust any factor: time of day, day of week, meeting density, etc.
  • Example: “Avoid Fridays” → Friday slots get -30% penalty

Real-World Examples

Startup Founder

Software Engineer

Executive Assistant

Error Handling

Invalid Time Format

Solution: Use 24-hour format like “09:00” or “17:00”

Invalid Timezone

Solution: Use IANA timezone database names like “America/Los_Angeles”

Conflicting Preferences

Solution: Ensure focus blocks are within work hours

Best Practices

1. Update Incrementally

2. Use Descriptive Scheduling Context

3. Set Realistic Limits

4. Align Focus Blocks with Energy

find_mutual_availability

See how preferences affect availability

check_availability

View current preferences

schedule_meeting

Schedule with updated preferences