Connect your event data to search, SQL, notebooks, and AI
One home for all your streaming data. Plug in the systems you already run, and every event you've ever produced becomes instantly usable. Nothing moves, nothing gets copied.
A real deployment with live data, in your browser. Free, and nothing to install.
Your operational tables
Everything, in one place
Your analytics tables
Your operational databases on one side, your analytics engines on the other. No pipelines in between. The plumbing stays invisible, the way it should be.
Works with what you already run.
No proprietary formats, no lock-in, no "supported integrations" page. Streambased connects the databases that run your business to the engines that answer your questions, through open standards your stack already uses. Compatibility is the product.
A packaged product, not a pile of infrastructure
You don't operate an engine. You open an experience. Everything below works on live streams and full history together, out of the box.
Data Explorer
Search every event you've ever produced, across every topic, every table, and all of history.
Live SQL
Plain SQL over one dataset: SELECT * FROM customers. Set HOT, COLD, or UNIFIED to query just the stream, just the archive, or span everything automatically.
Notebooks
Open a Jupyter notebook right on your data. Explore, chart, and model without exporting a single file.
Time Machine
"What did this customer look like yesterday at 3pm?" Rewind your data to any moment and look around.
Customer Timeline
Every event for a customer, in order, in one place. From first click to latest transaction.
Debugger
Replay and inspect event streams to find exactly what happened, and when, and why.
Connect once. Query forever.
No new storage system to learn. No architecture diagrams to study. Three steps, and every event you own is at your fingertips.
Connect
Point Streambased at the systems your data already lives in. No migration, no copies, no pipelines. Everything stays exactly where it is.
Discover
Every topic, schema, and table is detected automatically and appears in one searchable catalog. Nothing to configure, nothing to sync.
Use
Search, query, notebook, replay, build. Streaming and historical data behave like one dataset, because now they are.
Don't move your data. Connect it.
If your data already has a standard format, it shouldn't need a pipeline to become useful. Kafka and Iceberg are first-class citizens today, and the philosophy extends to every open standard your data speaks tomorrow.
- Don't copy your data into yet another system.
- Don't transform it unless you need to.
- Don't build pipelines just to make it usable.
Zero copies. Zero pipelines. Zero lag. Your data stays in the systems you already trust, and Streambased makes it all usable from one place.
See what disappears
The usual path from an operational database to your analytics stack is a chain of systems. Streambased collapses it to a single hop.
Every hop is another system to run, another copy to keep in sync, another place it can break.
Try it right now
A live playground, not a demo video
The playground is a real Streambased deployment with real data flowing live and resting in years of history. Search it, query it with SQL, or open a Jupyter notebook and explore, all in your browser.
- Live topics and historical tables, pre-loaded and queryable
- Full SQL editor where one SET flips between HOT, COLD, and UNIFIED
- One-click Jupyter notebooks running against the same data
SET mode = UNIFIED;
SELECT * FROM customers
WHERE region = 'EU';
Same query, same table. Flip mode and Streambased reads the stream, the archive, or both — no kafka. or iceberg. prefixes, no second system.
Query it yourself
Plain SQL over one dataset
Read-only, running right in your browser. Switch HOT, COLD, or UNIFIED to query just the stream, just the archive, or span everything.
| id | ts | amount | status |
|---|---|---|---|
| o_9105 | 2026-07-22 13:59:12 | 205 | failed |
| o_9104 | 2026-07-22 13:58:40 | 312 | failed |
| o_9103 | 2026-07-22 13:57:05 | 27 | ok |
| o_9102 | 2026-07-22 13:55:22 | 58 | ok |
| o_9101 | 2026-07-22 13:52:01 | 142 | ok |
| o_8006 | 2026-05-30 09:14:00 | 512 | ok |
| o_8004 | 2026-03-11 16:40:12 | 233 | refunded |
| o_8001 | 2025-12-02 11:20:33 | 410 | ok |
| o_8003 | 2025-09-18 08:05:50 | 178 | ok |
| o_8002 | 2025-06-27 14:33:10 | 96 | ok |
| o_8005 | 2024-11-14 19:22:41 | 44 | ok |
| o_8007 | 2023-08-01 07:10:05 | 61 | failed |
Where this goes
"The operating system for event data."
Nobody thinks "I need a filesystem supporting ext4." They think "I need somewhere to put my files." You shouldn't need a unified Kafka/Iceberg execution engine. You need somewhere all your event data lives.
Give your streaming data a home
Open the playground and run your first query on live data. Then connect your own clusters when you're ready.