QandA solution with LLM or Kendra

Kendra is the good starting point to build your QandA chatbot. Using Amazon Web Services (AWS) Kendra, you can build a Q&A chatbot that leverages natural language understanding and machine learning to deliver accurate answers to questions while keeping your data in your AWS account. If your data is already in AWS, you can config different source for your Kendra's index such as using S3 - https://docs.aws.amazon.com/kendra/latest/dg/data-source-s3.html

As a solution, usually you need to config a 'front-end' for your solution either via Lex service or a traditional Web based solution like API Gateway/Lambda and Website to handle the user interaction. You can also use other AWS services like Polly to turn the text into voice.

• https://aws.amazon.com/blogs/machine-learning/smarter-faq-bots-with-amazon-kendra/
• https://aws.amazon.com/solutions/implementations/qnabot-on-aws/

You can definitely combine the LLM solutions using AWS Sagemaker JumpStart Foundation Models(in preview mode and in selected regions) as your generative AI source with Kendra/OpenSearch as your KnowledgeBase. We provides LLM models for Text Summerization and Question Answering

Here are some links to start with

• https://aws.amazon.com/blogs/machine-learning/text-summarization-with-amazon-sagemaker-and-hugging-face/
• https://aws.amazon.com/blogs/machine-learning/aws-and-hugging-face-collaborate-to-make-generative-ai-more-accessible-and-cost-efficient/

Hope this is a good starting point for your Q&A chatbot solution

If you think this helps your solution design, please accept the answer