What is a minimum viable artificial intelligence solution?
Definition
A minimum viable artificial intelligence solution focuses on one high-leverage intervention rather than automating the complete process. It includes enough integration, evaluation and safeguards to produce evidence about usefulness with limited complexity and risk.
The solution should be tested on representative examples and with real users before its scope expands. Learning from a narrow deployment is more informative than adding features that have not been tied to a measured bottleneck.
Acronyms and aliases
minimum viable AI solution acronymsmallest responsible AI solution acronym
Related terms
Frequently asked questions
How small should an initial AI solution be?
It should address one measurable leverage point with the minimum data, authority and integration needed to test the expected operational benefit.
What makes a minimum AI solution responsible?
It has bounded permissions, representative evaluation, clear exception handling, user observation and a safe path to stop or correct failures.
