Notebook Entry 007
AI Capacity Estimation Ignores the Tools
April 6, 2026
Is capacity advice calibrated to what the tools now handle, or to doing everything by hand?
What happened
Amanda flagged that coaching advice was calibrated to her manual capacity — as if she were doing everything by hand. But LinkedIn posting was automated via Cowork, research was delegated to deep research, and building happened through Claude Code. The AI coach was being overly conservative about what Amanda could take on.
Root cause
Same error pattern as Claude Code estimating "8 hours" for a 10-minute task. The AI model was estimating effort without accounting for the tools available.
What changed
Five-step AI delegation checklist added to operating instructions. Every task filtered through: "Can this be done digitally?" Only tasks requiring physical presence or non-automatable manual input are truly manual. Planning now accounts for post-automation effort, not pre-automation totals.
The lesson
When you add AI tools to a workflow, the capacity model changes. Anyone advising on workload — human or AI — must evaluate what the tools handle, not what the tasks would cost manually. This is the exact same mistake orgs make when they adopt AI but don't update their headcount/timeline assumptions.
One implementation insight every other Tuesday.
What actually works when you ship AI in real organizations. Nothing you could get from a press release.