Chapter 01

The amnesia tax

Every conversation starts cold. Every tool starts cold. Every person starts cold.

The idea

Every conversation an AI has with your business starts cold.

  • Open a chat. Re-explain who you are, what your team is doing, why it matters. Five paragraphs in before the work starts.
  • Open the next chat. Same five paragraphs. Different person. Same context loss.
  • Switch tools. Same again. The thread is gone.
  • Multiply by every employee, every conversation, every tool. That is the amnesia tax.
  • It looks small per session. It compounds across the company.
  • And it is paid in the most expensive currency you have: senior people re-typing things they already wrote.
AI today is brilliant and amnesiac. The brilliance is borrowed. The amnesia is permanent.
In motion

Five sessions in a week. Same context paid five times. The work is the sliver in colour.

one week · five sessions · same context paid five timescontextwho we arewhat we are doingwhy it matterslast decisionpreferences~80% of sessionactual workAPI designMon · 09:14Engineer A · GPTcontextwho we arewhat we are doingwhy it matterslast decisionpreferences~80% of sessionactual workSame APIMon · 14:02Engineer B · Claudecontextwho we arewhat we are doingwhy it matterslast decisionpreferences~80% of sessionactual workRoadmapTue · 10:30PM · GPTcontextwho we arewhat we are doingwhy it matterslast decisionpreferences~80% of sessionactual workAPI againWed · 11:45Engineer A · Claudecontextwho we arewhat we are doingwhy it matterslast decisionpreferences~80% of sessionactual workSame flowThu · 16:20Designer · CursorThe amnesia tax compoundspaid in senior people re-typing what the company already knows
Three angles

The same step, depending where you sit.

Every prompt re-pays the context cost in tokens. Caching helps within a session. Across sessions and across people the cost resets to zero each time. The cache window is the visible part. The continuity gap is the invisible part.