Token Poker

Token Poker is a sprint planning tool for modern software teams that need to estimate more than just effort. Instead of only voting on story points, teams can also estimate the expected AI token usage for each task in the same shared planning room. This makes it especially useful for teams building AI-powered products, LLM features, agents, copilots, chatbots, or automation workflows where token consumption directly affects cost, scalability, and technical planning. With Token Poker, a team can discuss a user story, vote on its complexity, and at the same time forecast how many AI tokens the implementation or runtime behavior may require. This helps product managers, developers, and technical leads align early on both delivery effort and AI resource impact. The goal is to bring classic agile estimation into the AI era. Traditional sprint poker works well for understanding development complexity, but AI-driven applications introduce a new planning dimension: model usage, token costs, prompt size, context length, inference frequency, and expected user volume. Token Poker combines these concerns in one lightweight workflow, so teams can make better planning decisions before implementation starts. It helps teams answer questions like: * How complex is this story from a development perspective? * How much AI usage might this feature generate? * Which stories may become expensive to run at scale? * Are we underestimating the operational cost of an AI feature? * Should we simplify the prompt, reduce context, or rethink the architecture? By estimating story points and token usage side by side, Token Poker gives teams a clearer view of both engineering effort and AI resource demand. This makes sprint planning more transparent, more practical, and better suited for products where AI is part of the core system.