What is RAG Engine?
RAG Engine facilitates the connection of various external data sources to retrieval engines for Large Language Models (LLMs). Users can provide data through website URLs, files, Notion documents, or Slack channels via a single API call. The platform handles the necessary data processing steps, including parsing, chunking, cleaning, and embedding, eliminating the need for manual setup.
It incorporates a built-in, managed vector database hosted at cost, removing infrastructure management complexities for the user. RAG Engine enables unified search across all connected data sources, allowing applications to retrieve relevant data snippets instantly for use with LLMs. The service also includes scheduled data syncing to keep information current.
It incorporates a built-in, managed vector database hosted at cost, removing infrastructure management complexities for the user. RAG Engine enables unified search across all connected data sources, allowing applications to retrieve relevant data snippets instantly for use with LLMs. The service also includes scheduled data syncing to keep information current.
Features
- Data Ingestion: Connect various data sources like websites, files, Notion docs, and Slack channels with a single API call.
- Automated Data Processing: Built-in parsing, chunking, cleaning, and embedding of ingested data.
- Managed Vector Database: Hosted and maintained vector database provided at raw infrastructure cost.
- Unified Search: Retrieve relevant data snippets instantly from all connected sources.
- Scheduled Data Syncing: Keeps the connected data sources updated automatically.
Use Cases
- Enhancing LLM responses with up-to-date external information.
- Building AI applications that require access to specific documents or website content.
- Creating custom chatbots knowledgeable about specific company data or product documentation.
- Developing internal knowledge base search tools powered by AI.
- Simplifying Retrieval-Augmented Generation (RAG) pipelines for AI developers.
FAQs
-
Do I need to manage any infrastructure on my own?
Absolutely not. RAG Engine hosts everything from the ingestion pipeline to the vector database. You just make a couple of API calls. -
How does pricing work exactly?
Pricing has three components: RAG Engine Fee ($4.99/month, first month free), Vector Database (starting at $4/month for a DigitalOcean droplet, billed at cost), and Embeddings (OpenAI rates passed through with no markup, $0.010 to $0.065 per 1M tokens). -
What happens if my usage grows? How do costs scale?
If you need more storage than the default 512MB droplet, RAG Engine will scale your hosting to a higher-tier instance, still billed at cost. -
Which data sources will you support?
The first version focuses on website URLs and different file formats. Plans include adding integrations for Notion, Google Docs, etc., based on user feedback.
