Skip to content

Latest commit

 

History

History
73 lines (43 loc) · 2.87 KB

README.md

File metadata and controls

73 lines (43 loc) · 2.87 KB

📜 AWS-WHATSAPP-CHATBOT 📜

Latest CI/CD Action workflow

My Serverless WhatsApp chatbot on AWS, serving as a personal assistant with access to my private data.

  • It is able to assist me with my private data related to Events/Contacts/TODOs/Projects/Tips/etc...

Feel free to clone it and extend it to your own company information/data (at your own risk/responsibility)

Architecture 📝


How is the Generative-AI approach implemented on top of AWS?

  • RAG on top of Bedrock Knowledge Bases leveraging an OpenSearch Serverless Vector DB from PDF files.
  • Bedrock Agents to enable APIs and Database requests to fetch live data as part of the chain-of-thought process.
  • State Machine for different workflow's processing based on the user's input (text, voice-message, etc).

State Machine Process 🍂

The processing of messages is powered by an AWS Step Function that has multiple tasks based on the user's input:


Results (WhatsApp Assistant Demo) 🔮

Manual Steps (Only Once) ✋

WhatsApp Configurations

These steps show the creation of the "Meta Projects" and settings that will allow us to use WhatsApp Business APIs:

AWS Configurations

These steps show the creation of a Secret on AWS that will contain the required tokens/credentials for connecting AWS and Meta APIs.

Shoutouts 🙌

Thanks for all the inspiration and guidance on the Generative AI journey:

  • elizabethfuentes12 -> Gracias, Eli por inspirarme a ser un Developer Advocate!
  • micheldirk -> Thanks Michel for the inspiration for the low-level CDK constructs!

Author 🎹

Santiago Garcia Arango

Curious Solutions Architect experienced in DevOps and passionate about advanced cloud-based solutions and deployments in AWS. I am convinced that today's greatest challenges must be solved by people that love what they do.

LICENSE

Copyright 2024 Santiago Garcia Arango.