What problem does it solve?
Designing a retrieval-augmented generation (RAG) enterprise search system on Google Cloud requires coordinating many decisions: vector database selection, embedding pipelines, model serving, security, and deployment. This Skill guides you through a structured workflow that discovers requirements, produces a validated architecture, and delivers deployment-ready guidance.
Core Features & Use Cases
- Requirements Discovery: Systematically gathers functional and non-functional requirements (security, reliability, cost, performance, sustainability) and resolves ambiguities before any design work begins.
- Solution Architecture Generation: Produces product recommendations, Mermaid architecture diagrams, architecture descriptions, and design best practices grounded in official Google Cloud documentation.
- Validation and Packaging: Creates a validation plan with executable checks (terraform plan, vector index verification, latency tests) and compiles everything into a solution-architecture-guide.md document.
- Use Case: An enterprise wants conversational search over private documents using an open model (Gemma) with vLLM on GKE and AlloyDB as the vector store. The Skill walks them from requirements through Terraform deployment code.
Quick Start
Ask the agent to design a RAG-based enterprise search solution on Google Cloud using GKE and AlloyDB for your private document corpus.
