The challenge
Routine infrastructure requests were bottlenecked on the few people comfortable with the AWS CLI and console.
The client wanted non-technical staff to self-serve, without any risk of an AI accidentally changing or deleting production resources.
I’ll create this resource. Please confirm:
- Bucket
- invoices-archive-eu
- Region
- eu-west-1
- Access
- Private, public access blocked
- Versioning
- Enabled
boto3 call runs only after an explicit yes · claude 3.5 sonnet on bedrock
Architecture · 5 stages
- 01
Ask
Chat interface (Flask)
- 02
Understand
Claude 3.5 Sonnet on Bedrock
- 03
Validate
Collect required parameters
- 04
Confirm
Explicit yes/no approval
- 05
Execute
boto3 → AWS APIs
The solution
I built a conversational agent on Amazon Bedrock (Claude 3.5 Sonnet) with LangChain memory that turns natural-language requests into boto3 operations.
Every resource-changing action goes through mandatory field validation and an explicit yes/no confirmation before it executes. The agent proposes and a human decides. Bedrock session management keeps multi-turn context intact.
Engineering highlights
- Plain-language create, describe, update and delete for AWS resources
- Mandatory field validation before any action
- Explicit confirmation gate on every mutating operation
- Conversation memory across multi-step requests
Technology
- Amazon Bedrock
- Claude 3.5 Sonnet
- LangChain
- Python
- Flask
- boto3
- AWS
Related services
Building something similar?
Get a candid view of the architecture, risks and timeline on a free 30-minute call, or ask the AI assistant how this approach would map to your project.