Jasmin
    All Architectures
    AI SystemsIntermediate

    RAG Chatbot System

    A Retrieval-Augmented Generation (RAG) system that grounds LLM answers in your private data — PDFs, docs, databases, and knowledge bases. Users ask questions in natural language and receive accurate, cited answers from your content, not the model's training data.

    500 – 20Kusers supported
    $200 – $1,500/month infrastructure

    Architecture Diagram

    Interactive — hover over any node to see its role and description.

    UserChat UIReactFastAPIREST + SSERAG EngineLangChainLLMGPT-4o / ClaudeVector DBPineconeDoc StoreAWS S3PostgreSQLRedis CacheUserFrontendAPIBackendAI / MLDatabaseStorageCache

    Use Cases

    Internal knowledge base assistant for teams
    Customer support bot trained on product documentation
    Legal or compliance assistant over contract libraries
    Medical knowledge assistant over clinical guidelines
    Sales assistant grounded in product catalogs and pricing

    Technology Stack

    frontend

    ReactNext.jsTailwind CSS

    backend

    PythonFastAPILangChain

    database

    PostgreSQLPineconeRedis

    infrastructure

    AWS EC2AWS S3Docker

    ai

    OpenAI EmbeddingsGPT-4oClaude claude-sonnet-4-6Reranker

    Scalability Roadmap

    Stage 10 – 500 users· Single server + Pinecone Starter

    Single FastAPI server. Pinecone Starter plan (~1M vectors). Suitable for internal teams and early testing.

    Stage 2500 – 5K users· EC2 Auto Scaling + Pinecone Standard

    Multiple FastAPI instances behind a load balancer. Redis for cache. Async document ingestion workers.

    Stage 35K – 50K users· ECS + Managed Redis + Aurora

    Containerised on ECS. Aurora PostgreSQL. ElastiCache Redis cluster. Parallel ingestion pipeline.

    Stage 450K+ users· Multi-region + Pinecone Enterprise

    Regional deployments with data residency. Fine-tuned embedding models. Dedicated vector index per customer.

    Cost Breakdown

    Development Cost

    $8,000 – $20,000 (4–10 weeks)

    Infrastructure Cost

    $200 – $1,500/month (LLM API usage is the main variable cost)

    Maintenance Cost

    $1,000 – $3,000/month for model management and document pipeline upkeep

    Security Considerations

    Document access control — users only retrieve from their authorised namespaces
    PII scrubbing before documents are chunked and embedded
    LLM prompt injection prevention via input validation and sandboxed prompts
    All S3 documents encrypted at rest with customer-managed KMS keys
    Query logging with anonymisation for compliance and audit trails

    More Architectures

    AI Systems

    AI Agent Platform

    Autonomous AI agents with memory, tools & multi-step reasoning

    SaaS Platforms

    Multi-Tenant SaaS

    Scalable SaaS platform with per-tenant isolation and billing

    Marketplaces

    Marketplace Platform

    Two-sided marketplace with payments, reviews, and seller dashboards

    Need This Architecture Built?

    Get a detailed architecture plan, technology recommendations, development roadmap, and infrastructure estimation for your project.