Build a Streamhouse on Confluent Cloud
A Streamhouse is a category of data architecture, like a lakehouse, that keeps one current, trusted picture of your business for every consumer to read from. A consumer can be a person, a service, or an AI agent.
A Streamhouse captures change from every source into one current, trusted picture of your business, and every consumer reads from that picture.
On Confluent Cloud, you assemble a Streamhouse from the capabilities you already use, such as streaming, connectors, stream processing, governance, and queries, rather than buying it as a product.
These properties define a Streamhouse:
Real-time: State is continuously maintained as events arrive, rather than rebuilt on a batch schedule.
Always-on: A Streamhouse runs the business, so it’s engineered to production service levels rather than to reporting schedules.
Decentralized: A change is relayed everywhere it’s needed, in each consumer’s own interface, instead of forcing everyone to a single database.
Solutions you can build with a Streamhouse
A Streamhouse powers eight solutions, in two domains.
The first domain is the data and trust layer: a Streamhouse builds and protects one trusted picture of your business.
The second domain is the consumption layer: the Streamhouse turns this trusted picture of your business into outcomes.
Capture and maintain your business
Build and protect one current, trusted picture of your business:
Synchronize a change everywhere it is used: Capture each change once and deliver the same update to every dependent system.
Maintain one consistent state across participants: Keep one current value per key that every reader sees the same way, through processing changes and maintenance.
Isolate service evolution and failure recovery: Introduce a change or recover from a localized data failure in isolation, without destabilizing other teams.
Demonstrate data control without a parallel system: Prove access, lineage, and encryption from the system already running the business.
Run your business
Turn current state into outcomes:
Monitor the business to prevent negative impact: Recognize a developing failure or malicious act in the live stream and intervene early.
Decide across the last five seconds and the last five years: Make one decision across live conditions and deep history.
Ground and continuously adapt an AI agent: Serve live business context to agents and keep their logic current.
Route a computed outcome to its point of action: Deliver a produced decision straight to the system or person that acts on it.
Where you start: one always-current table
The entry point into a Streamhouse is a single continuously maintained table that you define once and serve to every consumer. To learn how you build and serve it, see Start here: the always-current table.
Next steps
To build a working Streamhouse end to end in about 30 minutes, see Quick Start: Build a Streamhouse on Confluent Cloud.
To provision a Streamhouse from your own PostgreSQL database in Confluent Cloud Console, see Build a Streamhouse with a Blueprint on Confluent Cloud.
To learn how the always-current table works and how you read from it, see Start here: the always-current table.
To capture change from your source systems into the Streamhouse, see Synchronize a change everywhere it is used.
If you’re new to Confluent Cloud, start with Quick Start for Confluent Cloud.