I Built AI Agents That Write Terraform for AWS (CrewAI + DevOps)

Description

What if AI agents could write your Terraform code, review it, and deploy AWS infrastructure for you? In this video, I build a multi-agent system using CrewAI that does exactly that — turning plain English requirements into production-ready Terraform and provisioning real AWS resources. This is agentic DevOps in action. If you're an AI/ML engineer curious about real-world agent applications, or a DevOps engineer wondering where LLMs actually fit in your workflow, this one's for you. 🎯 What you'll learn: - Designing a crew of agents for IaC: architect, coder, reviewer, deployer - Giving agents the right tools (file I/O, terraform CLI, AWS SDK) - Prompt engineering for Terraform generation that actually works - Guardrails: validation, terraform plan review before apply - Handling state, secrets, and cost control in an agentic workflow - Where this breaks — and when NOT to use agents for infra 🔗 Resources: - Code repo: https://github.com/andalike - CrewAI docs: https://docs.crewai.com - Terraform AWS Provider: https://registry.terraform.io/providers/hashicorp/aws 🛠️ Stack: Python 3.10+, CrewAI, Terraform 1.x, AWS (VPC/EC2/S3), OpenAI or Anthropic API ⚠️ Heads up: This deploys real AWS resources. Run `terraform destroy` at the end to avoid charges. Always review `terraform plan` output before letting agents apply. 👇 Would you trust AI agents with your production infra? Drop your thoughts below. #CrewAI #Terraform #AWS #AIAgents #DevOps