LONDON — Information streaming supplier Confluent at present introduced new capabilities in Confluent Intelligence and Confluent Cloud that streamline how real-time, synthetic intelligence (AI) functions are constructed and secured.
Confluent unifies the AI life cycle with instruments builders already stay in, integrating Apache Flink pipelines with dbt and introducing a totally managed Mannequin Context Protocol (MCP) server and Agent Expertise that permit AI handle streaming operations. Moreover, with automated personally identifiable info (PII) redaction and personal connectivity to exterior fashions by way of Azure Non-public Hyperlink, Confluent embeds enterprise-grade governance instantly into the information streams.
These updates take away the safety and complexity boundaries that cease organizations from shifting AI workloads into the actual world.
“Most AI initiatives fail earlier than they attain a single buyer as a result of the information layer breaks down,” stated Sean Falconer, Head of AI at Confluent. “Groups have the fashions and the mandate, however safety dangers and fragmented information cease them from transport. We’re fixing that by making the streaming layer the inspiration for safe, production-ready AI.”
The issue is widespread in keeping with a McKinsey report that finds “eight in ten corporations cite information limitations as a roadblock to scaling agentic AI.” Root causes are sometimes tied to safety groups blocking information from coming into AI pipelines over publicity dangers, and builders dropping hours to software switching to examine and handle the information streams their AI relies on. The result’s a sluggish, guide course of that turns what must be a quick iteration cycle right into a bottleneck.
The Engine for Safe, Scalable AI
Confluent Cloud and Confluent Intelligence kind the information streaming basis for production-ready AI that repeatedly processes historic and real-time information and delivers it as trusted context into AI functions. New capabilities add the safety controls and developer tooling that high-stakes industries require.
- Pure Language Operations: Builders can use a totally managed MCP server as a management airplane, permitting AI to construct, handle, and debug streaming operations utilizing pure language. Agent abilities add a second layer, encoding greatest practices and workflows so these operations are executed constantly and in step with organizational requirements. Collectively, they allow builders to create and repeatedly enhance real-time functions utilizing AI-powered instruments, bringing streaming into fashionable, agent-driven growth workflows. Usually accessible for Confluent Cloud.
- Automated Information Privateness: A brand new built-in ML perform for PII detection and redaction protects delicate info instantly in Flink SQL, with out customized code, exterior companies, or shifting information to a warehouse first. This unlocks extra AI use instances throughout extremely regulated industries like monetary companies, healthcare, and insurance coverage. Obtainable in early entry for Confluent Intelligence.
- Safe Connectivity: Assist for Azure Non-public Hyperlink ensures AI workloads keep off the general public web with safe, non-public paths to calling exterior fashions and querying exterior tables. Now, Flink jobs can securely hook up with Azure-hosted companies like Azure OpenAI, Azure SQL, and Cosmos DB over Microsoft’s non-public spine. Usually accessible on Confluent Cloud.
- Unified Engineering Workflows: The free, open-source dbt adapter brings Flink SQL on Confluent Cloud into dbt, the industry-standard framework information engineers use to construct and handle information pipelines. Groups can instantly outline, check, and deploy streaming pipelines utilizing the identical dbt instructions and venture construction they depend on at present. This lowers the barrier to Flink adoption and makes it simpler to increase present information workflows into real-time use instances. Usually accessible on Confluent Cloud.
- Flexibility with extra mannequin and vector database help: Confluent helps TimesFM fashions for strong anomaly detection in addition to Anthropic and Fireworks AI fashions, which builders can instantly use in Flink stream processing workflows to construct subtle real-time AI functions. Moreover, help for vector search on Amazon DynamoDB expands the fashionable AI stack ecosystem.
To be taught extra about at present’s bulletins, go to the Confluent blogs for Confluent Intelligence and Confluent Cloud. Highlights embrace the final availability of the Actual-Time Context Engine, which repeatedly delivers recent, ruled context for AI functions, and new totally managed connectors in Confluent Cloud that additional simplify information integration.
