With Retrieval Augmented Generation (RAG), external data is used to augment your prompts and can come from multiple data sources,
such as a document repositories, databases, or APIs. The first step is to convert your documents and any user queries into a
compatible format to perform relevancy search. To make the formats compatible, a document collection, or knowledge library,
and user-submitted queries are converted to numerical representations known as vector embeddings using embedding language models.
The original user prompt is then appended with relevant context from similar documents within the knowledge library.
This augmented prompt is then sent to the foundation model.
For this example, we created vector embeddings with Amazon Titan Model 'amazon.titan-embed-text-v1'
Sample Plan Document that was indexed can be viewed here
You can perform a RAG on the plan document via any of the following Bedrock models: AI21(Jurassic -2 Models), Anthropic’s Claude, Cohere
This demo can be used for a range of use cases including contact center agent support and mobile / web chatbots in order to increase the
ease in which members or call center agents can retrieve relevant information for member healthcare needs
What is the copay for a Specialist visit?
Are mental health services covered?
What is the maximum out of pocket for a family?
[What is the deductible for an individual?,What about family?]
[What is the deductible for an individual?,What about family?,What about prescription drugs?]
[What is the copay for Primary care visit?,What about specialist visit?,What about urgent care?]
Note: Enclosing question in [ ] makes it a conversation. Try it out!