Upload, query, and observe
Ephemeral Databases
Give agents the ability to create in-memory databases in milliseconds.OLAP, Top-k, Geospatial, and Vector Search in a single query.
Install with Homebrew
brew install hotdata-dev/tap/cliLoad SF Airbnb listings, search
hotdata auth register
hotdata databases create \
--name airbnb \
--table listings
hotdata databases load airbnb.listings \
--url https://hotdata.dev/data/sf-airbnb-listings.parquet
hotdata indexes create 'airbnb.listings[description]' \
--type bm25
hotdata search "cozy apartment with a view" \
--table airbnb.listings \
--select name,description
20×
Faster
than federated engines
90%
Lower Cost
than data warehouses
124+
Connectors
to databases, warehouses, lakes, SaaS, and more
Features
Secure Sandbox
Agents run in an isolated sandbox to protect production systems, with a complete audit trail of every action.
Unified Query API
A hybrid query interface—lookups, full-text search, OLAP, geospatial, and vector search in one path—across databases, data lakes, warehouses, and SaaS.
Millisecond Responses
Queries are cached and instantly reusable with no redundant computation or waiting.
Architecture Diagram
Fully managed or deployed in your own cloud environment.
Built with Rust, Apache Arrow, and Apache DataFusion
Questions?
Agents usually hit many sources at once with lots of concurrent queries—and they need answers quickly. Teams often end up gluing together connectors, caches, and routing by hand. Hotdata wraps that in one API so you get consistent, low-latency reads across your data without maintaining bespoke plumbing for every source.
They're fast, isolated places to query and experiment—one SQL surface across your backends—so agents can explore and join data without spinning up a new pipeline every time a source or workflow changes.
Agents rarely run one-and-done queries; they explore, branch, and retry. That pattern needs snappy responses and a consistent way to reach every system. Hotdata's sandboxes are built for that kind of back-and-forth instead of treating each source as a one-off integration.
Agents run in isolated environments with read-only access to your connected data—they can't change your systems of record. They can create temporary datasets for a session, scoped to the agent and safe to tear down, so exploration and iteration don't put production data at risk.
No. Hotdata is optimized for reads. We operate in read-only mode against snapshots of your databases, warehouses, and SaaS systems. Writes, updates, and deletes stay in each source system where they belong.
No. We only grant read access to connected sources, so agents can't remove or overwrite data at the source.
Hotdata layers on top of what you already run. Your databases, warehouses, and SaaS apps stay the source of truth; we add a low-latency path for agents and applications to query across them when they need to.
Apache DataFusion with Arrow-native execution.
Yes. With bring-your-own-cloud we deploy into your environment and operate the service—you keep control of infrastructure, access, and keys. If you'd rather not manage infra, we also offer a fully managed option.