RAG System Designer
Design and implement a Retrieval-Augmented Generation system
agentsragembeddingsvector-databaseadvanced
Prompt Template
Design a RAG system for: {use_case}
Data sources: {sources}
Query types: {queries}
Scale: {scale}
Provide:
1. Architecture overview with component diagram description
2. Embedding model recommendation
3. Vector database selection with reasoning
4. Chunking strategy
5. Retrieval pipeline (query -> context -> generation)
6. Evaluation metrics
7. Code skeleton for the core pipelineVariables
{use_case}Example: Internal documentation Q&A bot
{sources}Example: Confluence pages, PDF docs, Slack messages
{queries}Example: How-to questions, policy lookups, troubleshooting
{scale}Example: 10K documents, 100 queries/day
Example Output
## Architecture ``` User Query → Query Embedding → Vector Search → Context Assembly → LLM Generation → Response ```
Tips
- Consider hybrid search (semantic + keyword)
- Plan for document updates and re-indexing