AWS Bedrock Agents vs AWS Bedrock with Step Functions

Hi Amevy,

AWS Bedrock Agents are designed for conversational interactions, allowing you to build applications that can understand natural language inputs from users, interpret their intents, and provide responses based on a provided knowledge base. In your use case, a Bedrock Agent could directly interact with the user, understand their questions about building codes, provide answers from the knowledge base, and handle follow-up actions like creating tickets or generating application numbers based on the user's feedback.

AWS Step Functions, on the other hand, is a serverless workflow service that allows you to orchestrate and coordinate multiple AWS services and tasks into business processes. In your use case, Step Functions could be used to orchestrate the entire workflow, including:

1. Invoking the Bedrock model to generate an answer based on the user's input
2. Evaluating the user's feedback on the answer (e.g., through a separate user interface)
3. Based on the feedback, either creating a ticket for the knowledge base manager or generating a building application number
4. Returning the final result (either the ticket confirmation or the building application number) to the user

Using Step Functions would allow you to have more control over the entire workflow, including error handling, retries, and integrations with other AWS services or external systems. However, it would require you to manage the user interaction separately, potentially through a separate interface or API.

If your primary focus is on providing a natural, conversational user experience and the workflow is relatively straightforward, using Bedrock Agents alone might be the simpler approach.

Hope this helps