started · updated
Confluent introduces Streamhouse architecture for real-time AI data
Confluent, an IBM group company, has unveiled Streamhouse, a new architectural paradigm designed to provide AI agents, applications, and analytics systems with a continuous, governed view of real-time business context.
The architecture expands Confluent’s Data Streaming Platform to five pillars: Stream, Connect, Process, Govern, and the newly added Query. This fifth pillar is powered by Lightning Tables, a read-only serving layer on Confluent Cloud that is now generally available on Amazon Web Services. Lightning Tables allows applications to query live streaming data directly without requiring separate databases or caches. In internal testing, the technology achieved data latency from creation to serving of approximately 100 milliseconds, with query latency between 10 and 100 milliseconds and throughput exceeding 1,000 queries per second.
Streamhouse aims to address data fragmentation issues where AI agents might otherwise act on obsolete or conflicting information from heterogeneous systems. Confluent presented this development at its Current San Francisco 2026 conference, noting its relevance to rapidly scaling digital services in regions like Asia.