Synapse Engine

🧠 What Is the Synapse Engine?

The Synapse Engine continuously analyzes your knowledge graph and surfaces “Aha! moments” — meaningful connections between notes that you didn’t explicitly create.

It runs entirely on-device using text similarity analysis. No cloud, no external APIs. All processing happens locally, preserving your privacy.

🔍 Six Signal Detectors

The engine uses six different methods to find hidden connections.

1. Semantic Resonance

Finds notes that discuss similar ideas despite being written at different times. Uses Jaccard word-overlap similarity to identify notes with shared vocabulary and concepts. Only triggers for notes created more than 24 hours apart — closer notes are handled by Temporal Pattern detection instead.

2. Temporal Patterns

Identifies notes created within the same creative session (within 1 hour of each other) that aren’t yet linked. These are likely related ideas you had during the same train of thought.

3. Tag Convergence

Finds notes that share one or more tags but have no explicit connection. If you tagged two notes with “planning” but never linked them, Synapse flags it.

4. Contradiction Detection

Scans for notes containing opposing keyword signals — “should” vs “shouldn’t,” “for” vs “against,” “works” vs “fails,” etc. When two notes contain conflicting concept words, Synapse surfaces them for your review.

5. Chain Completion

Detects missing links in knowledge triangles. If note A connects to B, and B connects to C, but A doesn’t connect to C — Synapse suggests completing the triangle. Confidence is based on the strength of the existing edges.

6. Orphan Rescue

Finds isolated notes that have no connections at all, then checks if they’re textually similar to connected notes. If an orphan note shares significant content with a connected note, Synapse suggests linking them.

📋 Viewing Insights

Insights appear in the sidebar under the SYNAPSE section.

Insight Cards

Each insight shows as a card with:

• Type icon and color — each detector has its own icon and color (brain/purple for Semantic Resonance, clock/cyan for Temporal, etc.).
• Confidence badge — a percentage showing how strong the signal is.
• Node titles — the two notes involved, with their type icons.
• Description — a plain-English explanation of why this insight was surfaced.
• Connect / Dismiss buttons — tap Connect to create the relationship, or Dismiss to hide the insight.

Insight Header

The SYNAPSE section header shows a sparkle icon, the word “SYNAPSE,” and a badge with the count of active insights. When the engine is analyzing, a small spinner appears instead of the count.

⚙️ Managing the Engine

You control when and how Synapse runs.

Enable / Disable

In the sidebar SETTINGS section, toggle the Synapse Engine switch. When disabled, no insights are generated and the SYNAPSE section hides.

When It Runs

• On launch — a full analysis runs when you open Beatrice (if 5+ minutes have passed since the last run).
• After creating a note — a quick analysis checks the new note against existing notes.
• Cooldown — full analysis respects a 5-minute cooldown to avoid excessive processing.

Insight Limits

A maximum of 12 active insights are shown at once. Lower-confidence insights are automatically dismissed when the limit is reached. Dismissed insights are pruned after 7 days.

💡 Tips for Best Results

• Write substantive notes — Synapse needs at least ~10 words of content to analyze meaningfully.
• Use tags consistently — Tag Convergence only works when notes share tags.
• Don’t over-connect — The engine skips notes that are already linked, so a sparse graph yields more useful suggestions.
• Review contradictions carefully — The contradiction detector uses keyword heuristics. Not every match is a genuine conflict — use your judgment.