leaf

Identify context destruction in linear conversation streams.

• • Analyzed team deliberation logs
• • Mapped decision patterns
• • Identified derailment points

• • Focused on threading UI rather than structural context persistence

• • Linearity is the enemy of multi-threaded deliberation

• • Reframed as a non-linear graph deliberation system

Visualize non-linear conversation anchors.

• • Mapped segment-based anchoring logic
• • Identified context persistence nodes

• • Initial maps were too complex for mobile-first behavior

• • Navigation must feel linear even if the data structure isn't

• • Created a "thread-folding" UI logic

Test deliberation velocity in branching threads.

• • Ran side-by-side deliberation tests
• • Measured time-to-decision

• • Users lost track of "main" thread intent

• • Anchors must be explicitly visual and persistent

• • Integrated persistent segment headers for every branch

Build the branching graph engine.

• • Non-linear data store
• • Segment anchoring system
• • Graph navigation UI

• • Over-built the summarization features early on

• • Context persistence is more important than automated summary

• • Prioritized structural anchoring over AI synthesis

Calculate noise reduction and decision velocity.

• • Signal-to-noise ratio
• • Context retention time
• • Decisional speed

• • Metrics focused on message count, not signal quality

• • Signal-to-noise is a qualitative shift that requires behavioral tracking

• • Introduced focus-persistence tracking for deliberators