ai-pricing-model-picker
Installation
SKILL.md
Use when the user needs to pick or defend a pricing model for an AI agent product. Produces a recommended model, tier structure, and pricing anchors grounded in the budget the agent actually displaces.
Profile the agent first
Before recommending anything, get four answers — ask the user or infer from their product description, then confirm:
- Breadth of responsibility. Does the agent cover a comprehensive job function, or discrete tasks?
- Workload predictability. Consistent volume, or wildly variable?
- Attribution. Can a completed outcome be cleanly credited to the agent? Is autonomy high (agent finishes work alone) or low (copilot suggesting actions)?
- Budget displaced. Headcount, BPO/outsourcing spend, tools budget, or a performance line item? Headcount budgets run roughly 10x larger than tool budgets — this answer moves price more than any other.
Map to the model
| Model | Fits when | Watch out |
|---|---|---|
| Per agent (FTE replacement, fixed monthly fee) | Comprehensive function, consistent workload, headcount budget | Low differentiation; vulnerable to cheaper competitors |
| Per action (consumption) | Varied discrete tasks, unpredictable frequency, BPO budget | Highest commoditization risk — pricing pressure only moves down |
| Per workflow (completed multi-step sequences) | Multi-step processes with standardized intermediate deliverables | Standard workflows invite price compression; complex ones risk negative margin |
| Per outcome (pay for results) | High autonomy + high attribution, success metric the customer already tracks | Requires consistent performance; attribution disputes; bespoke outcomes proliferate contracts |