# Choosing & ROI frameworks
Source: https://humanspark.ai/frameworks/topic/choosing-and-roi/

Frameworks for deciding where AI is worth using and what it returns - scoring workflows, sizing the prize, and knowing when to stop. Eleven one-page models.

- **RATES** - A five-factor score that surfaces the highest-ROI tasks to automate first. (https://humanspark.ai/frameworks/rates/)
- **The Three Implementation Tiers** - Three ways to add AI to a process - very different cost, speed and risk. (https://humanspark.ai/frameworks/three-implementation-tiers/)
- **The €50k Threshold** - Only build custom if the problem costs more than this to solve by hand. (https://humanspark.ai/frameworks/fifty-k-threshold/)
- **The Four Pillars of Opportunity** - The four places AI creates value - to look past 'just save time'. (https://humanspark.ai/frameworks/four-pillars-of-opportunity/)
- **The Pilot Charter** - A one-page agreement that defines a pilot before any work begins. (https://humanspark.ai/frameworks/pilot-charter/)
- **The 30-Day Playbook** - A four-week structure that takes a pilot from target to go/no-go. (https://humanspark.ai/frameworks/thirty-day-playbook/)
- **Scale, Pivot or Kill** - The three-way verdict that ends a pilot before it can drift. (https://humanspark.ai/frameworks/scale-pivot-kill/)
- **The Golden Test Set** - Score the AI against known-good examples before launch, not opinions afterwards. (https://humanspark.ai/frameworks/golden-test-set/)
- **Quick Wins vs Big Wins** - Earn the big build with fast, cheap, internal wins first. (https://humanspark.ai/frameworks/quick-wins-big-wins/)
- **The Model Selection Matrix** - For boring text work the top models are roughly interchangeable. (https://humanspark.ai/frameworks/model-selection-matrix/)
- **Rules Before Guesses** - If a rule can do the job, use the rule - save AI's best guess for judgement, language and interpretation. (https://humanspark.ai/frameworks/rules-before-guesses/)