VegaRAG

VegaRAG is an open-source platform for AI chat agents that use retrieval-augmented generation with a user’s own data. It supports managed cloud use and self-hosting inside an AWS account, and the page describes it as a production-grade RAG platform.

Its documented capabilities include a live demo for RAG chat with document upload, SQL querying, and streaming responses. The platform also includes agent creation, configuration, and deployment, along with data ingestion from URLs, PDFs, CSVs, and plain text. A visual workflow studio is shown through a ReactFlow canvas, and the dashboard includes analytics charts, chat logs, deploy embed codes, system prompt settings, brand color controls, and per-agent chat UI branding with a title and logo. The chat UI service is described with thread history grouped by assistantId, an artifact renderer for structured outputs, and display of tool calls.

The architecture section names LangGraph StateGraph agents, Pinecone vector search, PostgreSQL Row-Level Security, Amazon Bedrock Nova, Microsoft Presidio for PII redaction, semantic caching, token bucket rate limiting, and dual-LLM hallucination checks. It also describes AWS Fargate deployment, AWS Cognito authentication, DynamoDB for agent and chat data, PostgreSQL for SQL analytics, OpenTelemetry, AWS X-Ray, and CloudWatch. The AWS setup includes Route 53 and ACM for a custom HTTPS domain, an Application Load Balancer, ECS Fargate microservices, and network isolation with VPC and security groups.

The page refers to users building AI chat agents with their own data, and it offers a managed cloud start-for-free option as well as self-hosted deployment. No pricing details beyond that are stated.