Dive Brief:
- IBM's Confluent rolled out a data architecture called a Streamhouse, which the company said “keeps business context current” for AI agents, according to a Thursday announcement.
- Confluent’s Data Streaming Platform unifies the ability to stream, connect, process, govern and now query data, serving as the basis for a Streamhouse architecture. The Lightning Tables layer on Confluent enables the query functionality for live data, giving “each agent and application a live, governed view of the business,” Shaun Clowes, general manager of Confluent and data at IBM, said in a press release.
- A Streamhouse architecture differs from the capabilities of data lakes and warehouses, which are typically used for reporting or batch jobs, to provide a more up-to-date view of the business. “The CIO’s decision shifts from where to move data to which business events agents need fresh and who authorizes agent access to them,” said Mitch Ashley, VP and practice lead, CIO and technology buyers, at The Futurum Group.
Dive Insight:
Enterprises deploying AI agents are running into a problem: data foundations that aren’t keeping pace.
While companies with strong data foundations can prepare the appropriate business context for agent operation and earn better returns on their AI investments, only 8% of enterprises describe their data as ready to support those capabilities, according to a Dun & Bradstreet survey of 10,000 businesses.
“The next era of enterprise AI will be won by the agents that know what’s happening in the business the moment they act,” Clowes said.
The Streamhouse architecture is an open category defined by a handful of companies, including Confluent, Ashley told CIO Dive in an email. The architecture carries changes out of existing systems, including mainframes, and delivers current state information to agents while data lakehouses maintain the history, he added.
“Enterprises spent the last decade building lakehouses that record what happened,” he said. “AI agents need to know what is true right now, and IBM is assembling a data estate that covers both the historical record and the current state of the business.”
Streamhouse architecture, supported by a handful of vendors including Confluent, is still in its early days, said Thornton Craig, senior director analyst at Gartner.
“Streamhouse offers a compelling narrative for management and delivery of complex real-time data,” Craig told CIO Dive in an email. “But widespread adoption and impact on building and scaling real-time architectures remains to be seen.”
Confluent, which IBM began acquiring late last year, isn’t the only vendor recognizing the importance of business context for agents.
Google added new capabilities to its Agentic Data Cloud, including a borderless Lakehouse to provide agents with a governed foundation for data across platforms such as Confluent, multicloud environments or on-premises, according to a Wednesday blog post. The tech giant is also previewing a self-learning enrichment capability in Knowledge Catalog, which is Google Cloud’s business context engine for agents.
“Knowledge Catalog learns from user intent and every human and agent interaction by continuously mining unstructured sources like Google Workspace, GitHub, Jira, Confluence, and SharePoint to close context gaps and evolve context automatically,” according to the blog post.
