1. Introduction
Building production-ready generative AI applications involves 3 stages:
- Ideation - rapid prototyping, testing apps with simple test prompts.
- Augmentation - evaluation, testing quality & safety with larger datasets.
- Operationalization - deploying apps and monitoring them for insights.
The challenge is in providing developers with a developer experience and tooling support to streamline these stages - without requiring a massive learning curve in new semantics, model-specific APIs or plaform-specific SDKs.
What if we could make the prompt the central unit of development - and build tooling to build agency and observability around it? That's the vision behind Prompty. Watch this Microsoft Build 2024 session for a deeper dive into end-to-end AI development using Prompty and AI Studio.
Using Prompty you can:
- Design Prompts - Create & refine prompt templates in an IDE
- Explore Models - Configure model parameters, diverse deployments
- Execute Prompts - Use built-in tools and custom runtimes to test prompts
- Observe Traces - Use built-in features to track and debug execution
- Compose Apps - Chain and compose Prompty assets for complex flows