Confluent Adds Query to Its Platform to Complete the Streamhouse Architecture

Data streaming vendor Confluent has introduced a new query capability to its platform, addressing a long-standing architectural gap by enabling enterprises to query the live state of a business directly from its streaming core. Announced during the Current 2026 conference in San Francisco, the update includes the general availability of Lightning Tables on AWS. This launch marks a significant step forward for the vendor, which is now an IBM company, as it seeks to eliminate the need for auxiliary downstream databases traditionally bolted onto streaming pipelines.

The Evolution of Enterprise Data Architectures

Historically, enterprise data management relied on batch processing to consolidate information from fragmented sources such as enterprise resource planning, customer relationship management, and billing systems. These legacy architectures typically copy operational data overnight into data warehouses or lakehouses. While suitable for human decision-makers reviewing daily reports, this batch-oriented approach creates latency that renders data days out of date by the time it reaches end-users.

Open-source technologies like Apache Kafka bypass batch processing by capturing individual changes as they occur and streaming that information directly to consuming systems. Confluent’s processing engine, Flink, further refines this approach by combining related events on the fly, such as matching an order event with a payment confirmation to establish a transaction status. However, applications previously lacked a native mechanism to query these processed results directly, forcing organizations to build and maintain separate operational databases at the end of the data stream.

Meeting the Demands of Continuous AI Agents

Industry leadership notes that the rise of autonomous artificial intelligence agents makes the latency of traditional batch architectures unsustainable. Unlike human workers who can identify and compensate for stale information, automated agents operating at machine speed require real-time accuracy. Relying on fragmented or outdated data can lead continuous enterprise agents to execute flawed decisions in rapid succession.

To mitigate this challenge, Lightning Tables maintain a constantly updated table of current business operations inside the Confluent platform. Internal testing indicates that these tables can serve fresh data approximately one hundred milliseconds after creation. Rather than consulting an overnight warehouse extract, applications and AI agents can query the live state of operations directly, ensuring consistency across customer-facing applications, gate displays, and mobile interfaces.

Core Components of the New Capability

The newly integrated Query capability unifies three distinct methods for extracting and utilizing data from the Confluent environment:

• Tableflow converts real-time data streams into open table formats, such as Iceberg and Delta, enabling analysis by standard lakehouse and warehouse tools.
• The Real-time Context Engine allows AI agents to request operational context through standardized protocols designed for connecting agents to enterprise tools.
• Lightning Tables provide a continuously updated internal repository that applications can query directly using standard SQL syntax or APIs.

By streamlining access to live operational data, organizations can reduce infrastructure overhead, eliminate redundant data copies, and remove the operational complexity of keeping multiple database systems synchronized. This architectural advancement positions data streaming platforms closer to the operational database and lakehouse territory, altering how enterprises capture and utilize real-time business intelligence.