Building Generative AI apps is hard enough when handling LLM peculiarities like hallucinations, latency, and cost. These apps can be increasingly more useful when integrating with enterprise and third-party APIs. However, rapid and scalable API integration with LLMs is even harder. GenAPI is on a mission to enable easy API use for Generative AI apps.
GenAPI has five goals to make this possible:
- Reusable functions library for popular use cases like checking the weather,
- Simple helper APIs to make app prototyping as easy as running cells on a notebook,
- Quickstart cookbook with notebooks for building Generative AI Apps using best practices,
- Comprehensive documentation with API, LLM, and app design tips, and
- Efficient workflow and project structure for building scalable Generative AI Apps.
We are currently exploring the following topics for addition to the GenAPI roadmap.
- https://github.com/openai/evals - can we use this to evaluate GenAPI apps?
- https://arxiv.org/pdf/2310.07343v1.pdf - can we use to enhance knowledge grounding for GenAPI apps.
- https://github.com/stanfordnlp/dspy - explore once docs are available.
- Please suggest more topics.