Using the AI Assistant

The AI assistant answers questions using your knowledge graph as context. It’s not a general chatbot — it’s your knowledge, amplified.

Effective Queries

Be Specific

Weak	Strong
”Tell me about learning"	"What techniques for improving long-term retention have I noted?"
"Help with React"	"What patterns have I documented for managing React state?"
"What do I know?"	"Summarize my notes on database optimization from the last month”

Trigger Synthesis

Use keywords that activate Sonnet for deeper analysis:

“Synthesize my notes on X”
“Compare my understanding of X vs Y”
“Analyze the trade-offs I’ve documented for X”
“Evaluate my decision to use X”

Ask About Connections

“How do my notes on memory relate to my notes on teaching?”
“What patterns emerge from my engineering decisions?”
“Are there contradictions in my notes about nutrition?”

What the AI Can Do

Recall: Find and surface relevant notes
Synthesize: Combine multiple notes into a coherent summary
Compare: Identify similarities and differences between note clusters
Challenge: Point out potential contradictions or gaps
Generate flashcards: Create review questions from notes

What the AI Cannot Do

Access the internet or external sources
Create notes on your behalf (you must write them)
Override your notes — it always cites your content as the source of truth
Access other users’ knowledge graphs

Understanding Source Citations

Every AI response includes source notes:


{
  "answer": "Based on your notes...",
  "sources": [
    { "id": "note_abc", "title": "...", "similarity": 0.91 },
    { "id": "note_def", "title": "...", "similarity": 0.78 }
  ]
}

similarity > 0.85: Strong match — the note directly addresses your question
similarity 0.70–0.85: Related context — pulled in via graph traversal
similarity < 0.70: Tangentially relevant — may or may not be useful