AI Infrastructure Agent: A Smarter Way to Manage AWS

AI Infrastructure Agent: A Smarter Way to Manage AWS

In modern DevOps, the focus is on speed and simplicity — reducing manual work and automating infrastructure. I built an AI-powered system that provisions AWS resources directly from natural language prompts, turning simple instructions into live cloud infrastructure.

For example:

“Create a t3.micro EC2 instance with Ubuntu 22.04.”

Within seconds, an EC2 instance is provisioned in AWS — fully configured, tagged, and ready to use.

No console. No YAML. No manual intervention.

Prerequisites

• AWS Account with appropriate permissions


• Docker installed on your machine


• AWS CLI configured


• AWS Bedrock access (we’ll set this up)

Step 1: Set Up AWS Bedrock Access

1.1 Navigate to AWS Bedrock Console

• Go to AWS Console → Search for "Bedrock"
  


• Click "Amazon Bedrock"
  

1.2 Request Model Access

• In the left sidebar, click Model access
  


• Click Request model access button


• Select any available text generation model (like Claude or Titan)


• Fill in required company details


• Submit the request

Step 2: Set Up AWS Credentials

Create AWS Access Keys

• AWS Console → IAM → Users
  


• Select your user → Security credentials
  


• Click Create access key
  


• Choose Command Line Interface (CLI)
  


• Download the credentials

Step 3: Run the AI Infrastructure Agent Container

Create Configuration Directory

mkdir ai-infrastructure-agent

cd ai-infrastructure-agent

Create Configuration File

mcp:

  server_name: "aws-infrastructure-server"

  version: "1.0.0"



aws:

  region: "us-east-1"



agent:

  provider: "OpenAI/Gemini/Anthropic/bedrock" # select your provider

  model: "Modelname"

  max_tokens: 4000

  temperature: 0.1

  dry_run: true

  auto_resolve_conflicts: false



web:

  port: 8080

  host: "0.0.0.0"

Create Environment File

# AWS Credentials

AWS_ACCESS_KEY_ID=your-access-key-here

AWS_SECRET_ACCESS_KEY=your-secret-key-here

AWS_DEFAULT_REGION=us-east-1

Run the Docker Container

docker run -d \

  --name ai-infrastructure-agent \

  -p 8080:8080 \

  -v $(pwd)/config.yaml:/app/config.yaml:ro \

  -v $(pwd)/states:/app/states \

  --env-file .env \

  ghcr.io/versuscontrol/ai-infrastructure-agent

Verify Container is Running

docker ps

docker logs ai-infrastructure-agent

Step 4: Access the Web Dashboard

4.1 Open Your Browser

Navigate to:

http://localhost:8080

4.2 Dashboard Overview

You should see the AI Infrastructure Agent dashboard featuring:

• Natural language input field


• Infrastructure state viewer


• Real-time monitoring


• Dependency graph visualization

Step 5: Test with Simple Commands

Create a t3.micro EC2 instance with Ubuntu 22.04

The AI shows:

• Resources to be created


• Estimated costs


• Dependencies


• Security configurations
  
  

Dry Run Mode

Since dry_run: true is enabled, no actual resources are created until you approve.

Disable dry_run mode and try to create a resource

Step 6: Cleanup Resources

Delete the resources created in the last session

Manual Cleanup

Optionally clean up via AWS Console or CLI using commands shown by the agent.

Step 7: Container Management

# Stop the container

docker stop ai-infrastructure-agent



# Start the container

docker start ai-infrastructure-agent



# View logs

docker logs -f ai-infrastructure-agent

Benefits

✅ Natural Language Interface – No AWS syntax needed

✅ Visual Monitoring – Real-time dependency graph

✅ Dry Run Mode – Prevents accidental creation

✅ Cost Awareness – Estimates before execution

✅ State Management – Tracks infrastructure changes

Conclusion

The AI Infrastructure Agent revolutionizes AWS infrastructure management by combining the power of AI with intuitive natural language commands. Whether you're a DevOps engineer, developer, or cloud architect, this tool simplifies complex infrastructure tasks while maintaining safety and visibility.

The containerized deployment ensures consistent behavior across different environments, while the web dashboard provides an intuitive interface for both beginners and experts.

Happy Learning

Prithiviraj Rengarajan

DevOps Engineer