Search

Search components, tokens, patterns, architecture

Patterns

Pattern vs Insight

A pattern is a relationship the system notices — neutral, no recommendation. An insight is a pattern that has meaning: it answers “so what?” and is backed by medium/high confidence. The stats row counts them separately, and the copy never blurs the line.

Sleep Playbook

What has sleep taught me?

Expand both rows — the confidence chip changes the language, not just the badge

Rules

  1. 01Pattern: neutral observation. “Brain fog often follows poor sleep.”
  2. 02Insight: meaning attached. “Sleep affects your clarity more than your energy.”
  3. 03Patterns found = all detected items, any confidence.
  4. 04Insights learned = medium/high-confidence items only.
  5. 05Eligibility: observed ≥ 3 times AND rate ≥ 60%. Confidence: Low 3–4 · Medium 5–8 · High 9+ observations.

Examples

Stats row

“42 days logged · 11 patterns found · 4 insights learned” — the gap between 11 and 4 is honesty made visible.

Anti-patterns

What breaks this pattern

  • Calling every detected correlation an insight.
  • Recommending off a 3-observation pattern.

Do

  • ·Show evidence counts in expanded rows
  • ·Let users see all patterns, weakly held ones included

Don't

  • ·Inflate the insight count
  • ·Attach actions to low-confidence patterns