Methods for discovering projects and linking related communications.

Project Identification Strategies

Keyword Clustering:

• Extract frequent noun phrases from calendar events
• Find recurring terms in email subjects
• Identify Slack channel names and topics
• Group by co-occurrence (terms appearing together)

Common Patterns:

• Project code names (alpha, beta, launch-2025)
• Feature names (new-dashboard, api-v2)
• Client/customer names
• Initiative names (redesign, migration, optimization)

Signal Strength:

• Calendar event title: High signal
• Email subject: Medium-high signal
• Slack channel name: High signal
• Message content: Lower signal (noisy)

Cross-Platform Mapping

Calendar → Project:

• Event title contains project keyword
• Recurring meetings suggest ongoing project
• Attendee overlap indicates project team

Email → Project:

• Subject line keywords
• Thread participants (team members)
• Date ranges align with project timeline
• Attachment names/types

Slack → Project:

• Channel name matches project
• Channel purpose/topic mentions project
• Message keywords and frequency
• Pinned messages with project info

Thread Association Rules

Strong Association (confidence >80%):

• Exact keyword match in title/subject
• Same participant set (>70% overlap)
• Temporal clustering (same week)
• Direct reference to other threads

Moderate Association (confidence 50-80%):

• Partial keyword match
• Some participant overlap (40-70%)
• Related time period (same month)
• Similar topics/tags

Weak Association (confidence 20-50%):

• Fuzzy keyword similarity
• Minimal participant overlap
• Loose temporal connection
• Suggested by user or context

Project Metadata Extraction

Timeline Construction:

project_timeline:
start_date: earliest event/message date
last_activity: most recent event/message
duration_days: calculated
activity_frequency: messages per week
phases: detected from calendar/keywords

Team Identification:

team_members:
name: Person A
role: inferred from messages/title
participation: message count
name: Person B
# ...

Status Inference:

• Active: Recent activity (<7 days)
• In Progress: Activity within 30 days
• Completed: No activity >60 days + completion keywords
• Paused: Gap in activity >30 days then resumed

Completion Keywords: "done", "completed", "shipped", "launched", "closed", "final"

Project Directory Structure

projects/
[project-slug]/
README.md         # Project overview and status
timeline.md       # Chronological activity
team.md          # Team members and roles
links.md         # References to emails/slack/calendar
notes.md         # User notes and context

README.md Template:

---
project: Project Name
status: active|paused|completed
start_date: YYYY-MM-DD
last_activity: YYYY-MM-DD
team_size: 5
platforms: [calendar, gmail, slack]
keywords: [keyword1, keyword2]
---
Overview
Brief description inferred from context
Recent Activity
Latest calendar event
Latest email thread
Latest slack discussion
Key Milestones
Detected from calendar and messages
Next Steps
Inferred from recent communications

Grouping Heuristics

Participant-Based:

• Messages with same 3+ participants likely related
• Calendar invites with overlapping attendees
• Slack threads in project-specific channels

Temporal-Based:

• Activity bursts indicate project phases
• Regular cadence (weekly standups) = active project
• Long gaps = project transitions or pauses

Keyword-Based:

• Co-occurring terms across platforms
• Consistent terminology usage
• Related technical terms or jargon

Auto-Detection vs User Input

Auto-Detect When:

• Strong keyword signals across 2+ platforms
• Clear participant clustering
• Obvious calendar event patterns
• Confidence score >70%

Prompt User When:

• Ambiguous groupings (confidence 50-70%)
• Multiple possible project names
• Overlapping teams (could be 2 projects)
• User can provide context/names

Output Format

links.md Structure:

# Project Links
Calendar Events
Event Title - Date
Event Title - Date
Email Threads
Thread Subject - Date
Thread Subject - Date
Slack Discussions
Channel Name - Date
Channel Name - Date

Maintenance Strategy

Incremental Updates:

• Add new threads to existing projects
• Update last_activity dates
• Append to timeline
• Preserve manual edits to README

Archival:

• Move completed projects to archive/
• Keep metadata but trim old links
• Maintain searchable index