RAGformation: AI-Powered Automated Cloud Configuration Generation

Cloud architecture design shouldn’t take weeks. We built RAGformation — an AI-powered system that automates cloud configuration, service selection, and cost estimation in minutes.

The Problem

Organizations face a complex challenge: selecting the right cloud services, designing scalable architectures, and estimating costs accurately. This process typically requires:

• Days or weeks of research and planning
• Deep expertise across multiple cloud providers
• Repeated manual cost calculations
• Architecture validation and refinement

What if we could automate all of this?

The Solution: RAGformation

RAGformation is a hackathon-winning tool that uses AI agents and retrieval-augmented generation (RAG) to intelligently design cloud architectures. Rather than relying on manual expertise, the system leverages AI to:

Core Capabilities

Automated Service Selection

• Analyzes natural language requirements
• Uses RAG with Pinecone vector store to retrieve relevant cloud services
• Recommends optimal services for your use case
• Supports AWS, Azure, and Google Cloud

Visual Architecture Design

• Generates architecture flow diagrams automatically
• Creates YAML-based visual representations
• Validates diagram correctness before output
• Produces publication-ready documentation

Intelligent Cost Estimation

• Integrates with AWS pricing APIs
• Provides accurate, detailed cost breakdowns
• Estimates total cost of ownership
• Suggests cost optimizations

Dynamic Refinement

• Adjusts recommendations based on feedback
• Respects budget constraints and requirements
• Iteratively improves suggestions
• Adapts to changing priorities

Technology Stack

LlamaIndex Agent Framework

• Orchestrates multiple specialized agents
• Manages complex multi-step workflows
• Enables seamless LLM integration

Vector Database (Pinecone)

• Stores scraped AWS architecture documentation
• Powers semantic search for service recommendations
• Enables fast, relevant retrieval at scale

Box Integration

• Organizes and manages architecture documentation
• Stores scraped cloud service data
• Maintains searchable knowledge base

OpenAI Integration

• Uses ChatGPT-compatible API standard
• Powers the conversational interface
• Drives all AI reasoning and decision-making

The Agent Workflow

RAGformation orchestrates six specialized agents working together:

1. Concierge Agent — Engages with users to understand requirements through natural conversation
2. RAG Agent — Retrieves relevant cloud services information from the knowledge base
3. Diagram Agent — Generates architecture diagrams in YAML format
4. Validation Agent — Checks diagram correctness and completeness
5. Pricing Agent — Calculates costs using real pricing data
6. Reporter Agent — Outputs the final comprehensive documentation package

Each agent specializes in its domain while the LlamaIndex framework orchestrates seamless handoffs between them.

Business Impact

RAGformation delivers measurable value:

Faster Deployment — From weeks to minutes of cloud architecture planning

Informed Decisions — Data-driven service selection based on actual requirements

Cost Optimization — Accurate pricing and cost breakdowns prevent surprises

Organizational Agility — Enable faster cloud adoption and experimentation

Open Source

The complete code is available on GitHub — bringing cloud configuration automation to everyone.

This is what happens when you combine AI agents, retrieval-augmented generation, and domain expertise. Not gatekeeping the future of cloud architecture, but building it in public.