Documentation
Use cases
What developers build on Spark: multimodal stores, in-app search, company brains, agent memory, and more.
Spark is smart object storage. You write files with the S3-compatible API, the HTTP API, or the CLI, and Spark reads and indexes them for search and query. These are the products teams build on top. Each one links to the guide that shows how.
For product teams
Multimodal store for a SaaS product
Store each customer's files — calls, video, documents, and JSON — behind one object API. Spark indexes every object for search and query, so you drop a second data store.
How to build: Store objects →For product teams
In-app search
Add fast keyword search over user content. Retire a separate search service and keep one dependency in your stack.
How to build: Keyword search →For media teams
Media and video library search
Search across video and audio by what is said or shown. Jump to the exact moment; each result cites a timestamp.
How to build: Vector / RAG search →For internal tools
Company brain
Index docs, wikis, and meeting notes. Ask one question across all of them and get an answer with sources.
How to build: Agentic query →For agent builders
Agent memory and knowledge
Give an agent durable memory and a shared knowledge base. Reach the same store over the API or the MCP server.
How to build: Agents & MCP →For platform teams
Multi-tenant isolation
Partition each customer's data on writes, search, and query. Keep one bucket per tenant boundary and never leak results across tenants.
How to build: Tenant isolation →Start building
Store your first object and read it back in a few minutes, then follow the guide for your use case.