The modern corporate landscape is dominated by microservices and APIs developed rapidly, creating a demand for efficient documentation practices due to limited development time.
Through an experiment combining OpenAPI with Retrieval Augmented Generation, I explored improving the generation and accessibility of API documentation using LLM and RAG technologies.
Retrieval Augmented Generation enhances response accuracy by integrating generative models with real-time data retrieval from external sources, such as databases or vector stores.
To implement a RAG system, the first step is to create a 'knowledge base' from the OpenAPI specification, breaking it into chunks for efficient retrieval.
[
Collection
]
[
|
...
]