Optimize AI-driven outcomes with prompt evaluations

Enabled for	Public preview	General availability
Admins, makers, marketers, or analysts, automatically	May 15, 2025	-

Business value

The prompt accuracy scoring feature in AI Builder’s prompt builder gives you data on how effective a prompt is. It offers a high level of testability and, more importantly, evaluates the results of the prompt. With this feature, you can find areas to improve and optimize the accuracy of your prompt, so AI-driven results better align with your business goals.

Feature details

The prompt accuracy scoring feature in AI Builder’s prompt builder helps you build a test suite and check your prompt performance across different versions of prompt development. With these detailed assessments, you can make informed decisions about using prompts in agents, apps, and flows, moving capabilities to production, and improving prompts. This feedback on the effectiveness of your AI prompts helps you optimize clarity and precision.

As you create or refine prompts, the feature analyzes the prompt structure, language, and relevance to the task. It assigns a confidence score to each test case prediction that shows the expected performance of the prompt. The score comes from factors such as specificity, complexity, alignment, and custom assertions. With this score, you get actionable insights to improve prompt phrasing or reduce ambiguity.

By giving you a clear, quantifiable measure of prompt quality, the accuracy scoring feature simplifies the prompt engineering process. It enhances model outcomes and reduces iteration time. With this feature, you get more efficient and reliable AI interactions across use cases.

Geographic areas

Visit the Explore Feature Geography report for Microsoft Azure areas where this feature is planned or available.

Language availability

Visit the Explore Feature Language report for information on this feature's availability.

Additional resources

Batch testing for prompts (preview) (docs)