Story points vs token estimates

Story points answer: how much effort does this ticket feel like for the team? They include complexity, risk, unknowns, testing effort, coordination, and implementation effort.

Token estimates answer: how much AI model usage might this ticket require? They include prompting, codebase context, AI-assisted debugging, test generation, documentation generation, code review assistance, and repeated model calls.

A ticket can be low in story points but high in token usage. For example, refactoring a legacy billing helper might be 3 story points but 1M+ tokens because the AI assistant may need many files, tests, logs, and repeated prompts.

A ticket can also be high in story points but low in token usage. Aligning product requirements with a legal team might be 8 story points but 100k tokens because the difficulty is coordination, not model usage.

Story points estimate team effort. Token estimates forecast expected AI usage. Token Poker lets teams estimate both.