Skip to main content

Data Pool

The Data Pool is the central interface for managing indexed data in SmythOS. It allows you to create Data Spaces, upload files or URLs, and optionally connect external vector storage using Custom Storage. This enables agents to use retrieval-augmented generation (RAG) and answer queries using relevant context from your documents.

What the Data Pool enables

Agents in SmythOS can use your files, pages, and structured content for precise responses. The Data Pool powers this by indexing data into searchable embeddings.

How the Data Pool supports RAG​

RAG stands for retrieval-augmented generation. When an agent is connected to a Data Space, it can search through indexed content and use the most relevant snippets to generate informed answers. This makes your agents more accurate, grounded, and useful for knowledge-based tasks.

RAG gives your agent real knowledge

Without data, your agent only knows what you prompt it with. With RAG, it retrieves meaningful information from your own sources before generating output.

Key features of the Data Pool​

FeatureDescriptionLearn More
Data SpacesContainers where you upload and manage files or URLs for indexingCreate Data Spaces
Custom StorageConnect Pinecone or another vector store to store embeddings externallySet Up Custom Storage
Indexing StatusSee whether each data source has been successfully indexed and is available for agent retrievalIncluded in the Data Spaces guide
Data source previewReview uploaded files or linked content from the interfaceManaged inside each Data Space
Use one Data Pool, many agents

You can reuse the same Data Spaces across multiple agents. This makes your content modular and efficient to maintain.

Using the Data Pool​

To access the Data Pool:

  1. Open the SmythOS Studio sidebar
  2. Click Data Pool
  3. View your list of existing Data Spaces
  4. Check the indexing status for each row
  5. Click any row to open the full Data Space view
Data Pool table in SmythOS

The Data Pool shows your content containers, index status, and linked storage at a glance.

Storage options​

The Data Pool supports both internal and external vector storage for embeddings.

Storage TypeHosted BySetup RequiredBest ForDocs
Internal StorageSmythOSNoneQuick setups, default RAG usageData Spaces
External Storage (Pinecone)You (Pinecone)API key and index infoFull control, enterprise use, scalable vectorsCustom Storage
Pinecone index requirement

Make sure your Pinecone index uses a vector dimension of 1536 or indexing will not work.

What's Next​