← Spaces

Breadcrumbs

Understanding breadcrumbs: the fundamental unit of knowledge in Bread.

What is a Breadcrumb?

A breadcrumb is a saved conversation or piece of knowledge that has been captured, enriched by AI, and indexed for future search and retrieval. Think of it as a marker on your trail - not just saving the content, but understanding and categorizing it.

Anatomy of a Breadcrumb

  1. 1
    Original Content

    The full conversation, notes, or discussion you saved

  2. 2
    AI Enrichment

    Auto-generated summary, category, key points, concepts, and action items

  3. 3
    Search Index

    Indexed for semantic search - find by meaning, not just keywords

  4. 4
    Metadata

    Context about client, session, trail, and more

When to Drop Breadcrumbs

Drop breadcrumbs at key moments to build a valuable knowledge trail:

✓ Good Times

  • • After solving a complex problem
  • • When making important decisions
  • • At the end of productive conversations
  • • When you learn something valuable
  • • Before context switching
  • • When sharing knowledge with the team

✗ Less Useful

  • • Very short exchanges without substance
  • • Repeated conversations about the same topic
  • • Pure small talk or greetings
  • • Test messages
  • • Conversations with no future value

Types of Content to Save

💬 Conversations

Code reviews, debugging sessions, feature planning, learning conversations

📋 Decisions

Architecture choices, technology selections, API design, trade-offs

🔬 Research

Library comparisons, performance investigations, best practices

🐛 Problem Solving

Bug analysis, solution approaches, lessons learned

✏️ Meeting Notes

Standup summaries, design reviews, planning outcomes

💡 Ideas

Feature ideas, optimization opportunities, innovation concepts

Breadcrumb Categories

AI automatically categorizes each breadcrumb into one of these types:

planning

Design & architecture

implementation

Coding & building

research

Learning & exploring

debugging

Troubleshooting

documentation

Writing docs

decision

Important choices

question

Q&A

general

Miscellaneous

Best Practices

  • • Drop breadcrumbs liberally: It's better to save too much than too little
  • • Use highlights: Call out the most important points for better AI enrichment
  • • Include context: Longer breadcrumbs give better AI summaries and embeddings
  • • Set appropriate scope: Use private for personal notes, trail for team knowledge
  • • Send metadata when available: Rich context enables better analytics
  • • Trust the AI: Categories and summaries are usually accurate