Note Creation Strictness

Control how aggressively Rowboat creates notes from emails

Rowboat’s knowledge graph system supports three strictness levels that control how aggressively notes are created from emails. This helps you manage signal vs. noise based on your email volume.

Meetings always create notes at all strictness levels. Strictness only affects email processing.

Configuration

Configuration File

Strictness is configured in:

~/.rowboat/config/note_creation.json

Default (High)

{
  "strictness": "high",
  "configured": false
}

Schema Fields

Field	Type	Required	Description
strictness	string	Yes	One of: "high", "medium", "low"
configured	boolean	Yes	Whether you’ve explicitly set this (false = using default)

Strictness Levels

High Strictness

Philosophy: “Meetings create notes. Emails enrich them.”

Best for: Users with high email volume who want minimal noise

Behavior:

• ✅ Meetings always create notes
• ✅ Emails can update existing notes (people you’ve already met)
• ✅ Warm intros from known contacts create notes
• ❌ Emails never create new notes (except warm intros)
• ❌ All other emails are skipped

Use cases:

• You get 100+ emails per day
• Most emails are newsletters, automated messages, or cold outreach
• You want your knowledge graph to focus on people you’ve actually met

Medium Strictness

Philosophy: “Both create notes, but emails require personalized content.”

Best for: Balanced capture of relevant business contacts

Behavior:

• ✅ Meetings always create notes
• ✅ Personalized business emails create notes
• ✅ Warm intros from anyone create notes
• ❌ Consumer services (Amazon, Netflix, banks) skipped
• ❌ Mass newsletters skipped
• ❌ Generic cold sales skipped
• ❌ Automated/system emails skipped

Use cases:

• You get 50-100 emails per day
• Mix of personal and business email
• You want to capture important business contacts automatically
• You don’t want noise from consumer services

Low Strictness

Philosophy: “Capture broadly. Never miss a potentially important contact.”

Best for: Users with low email volume who want comprehensive capture

Behavior:

• ✅ Meetings always create notes
• ✅ Any identifiable human sender creates notes
• ✅ Includes recruiters, sales reps, support staff
• ✅ Includes consumer service contacts
• ❌ Only obvious automated emails and newsletters are skipped

Use cases:

• You get fewer than 50 emails per day
• You want to capture every potential contact
• You prefer to have a note you don’t need rather than miss someone important

Decision Matrix

What gets captured at each strictness level?

Email Type	High	Medium	Low
Meetings	✅	✅	✅
Mass newsletters	❌	❌	❌
Automated/system emails	❌	❌	❌
Consumer services (Amazon, Netflix, banks)	❌	❌	✅
Generic cold sales	❌	❌	✅
Recruiters	❌	❌	✅
Support reps	❌	❌	✅
Personalized business emails	❌*	✅	✅
Warm intros (from known contacts)	✅	✅	✅

*At High strictness, personalized emails can only update existing notes, not create new ones.

Auto-Configuration

On first run, Rowboat analyzes your email patterns and recommends a strictness level:

1. Email Analysis

   Rowboat examines your recent emails to understand:

   • Total volume
   • Number of unique human senders
   • Percentage of consumer services vs. business emails
   • Patterns of cold outreach vs. ongoing conversations
2. Recommendation Logic

   Based on analysis:

   Condition	Recommended Level	Reason
   More than 100 human senders	High	Avoid overload
   50-100 senders	Medium	Balanced approach
   More than 50% consumer services	Medium	Filter noise
   Less than 30 senders	Low	Comprehensive capture is manageable

3. You Decide

   Rowboat presents the recommendation, but you make the final choice. Set configured: true once you’ve made your decision.

Choosing Your Strictness Level

Choose High if:

• ✅ You get 100+ emails per day
• ✅ Most emails are newsletters or automated
• ✅ You primarily meet people before email conversations
• ✅ You want your knowledge graph tightly focused

Choose Medium if:

• ✅ You get 50-100 emails per day
• ✅ Mix of business and personal email
• ✅ You want to capture relevant business contacts
• ✅ You don’t want clutter from consumer services

Choose Low if:

• ✅ You get fewer than 50 emails per day
• ✅ You want comprehensive contact capture
• ✅ You’d rather have extra notes than miss someone
• ✅ You can handle more notes in your knowledge graph

Examples

Example 1: Startup Founder (High Volume)

Email profile:

• 200+ emails/day
• Many newsletters, investor updates, cold pitches
• Regular meetings with team, investors, customers

Recommended: High strictness

Result: Knowledge graph focuses on people from meetings. Cold emails don’t create noise.

Example 2: Consultant (Medium Volume)

Email profile:

• 60-80 emails/day
• Mix of client work, proposals, networking
• Some cold outreach, occasional consumer emails

Recommended: Medium strictness

Result: Captures real business contacts, filters out noise. Warm intros create notes immediately.

Example 3: Individual Contributor (Low Volume)

Email profile:

• 20-30 emails/day
• Mostly internal team communication
• Occasional external contacts

Recommended: Low strictness

Result: Comprehensive capture of everyone you interact with. Easy to manage the volume.

Changing Strictness

You can change your strictness level at any time:

1. Edit Configuration

   Edit ~/.rowboat/config/note_creation.json:

   {
     "strictness": "medium",
     "configured": true
   }
   
2. Restart Rowboat

   Restart Rowboat to apply the new setting.

3. Existing Notes Unchanged

   Changing strictness only affects future email processing. Existing notes are not modified or deleted.

Technical Details

Prompt Files

Each strictness level uses a different agent prompt:

apps/x/packages/core/src/knowledge/
├── note_creation_high.md      # High strictness rules
├── note_creation_medium.md    # Medium strictness rules
└── note_creation_low.md       # Low strictness rules

Change Detection

Strictness affects which files get processed. The system uses a hybrid approach:

1. mtime check: Quick check if file modification time changed
2. Hash verification: If mtime changed, compute content hash
3. Process: Only if content actually changed

State tracked in:

~/.rowboat/knowledge_graph_state.json

Processing Flow

Troubleshooting

Too many notes being created

→ Increase strictness level (low → medium → high)

Missing important contacts

→ Decrease strictness level (high → medium → low)

Want to reprocess all emails

# Delete state file to force reprocessing
rm ~/.rowboat/knowledge_graph_state.json

Reprocessing all emails can take time and may create duplicate notes if not done carefully.

Best Practices

1. Start with recommended: Let Rowboat analyze your email and suggest a level
2. Adjust based on results: After a week, review your knowledge graph and adjust
3. Consider email habits: If you aggressively archive/delete, use lower strictness
4. Review periodically: Your email patterns may change over time
5. Reset if needed: You can always reset and reprocess with different settings

Related Configuration

• Models Configuration - Configure the LLM used for note creation
• Google Setup - Connect Gmail to enable email processing