Confluent, an IBM company, has released new capabilities in Confluent Intelligence and Confluent Cloud aimed at removing the security and complexity barriers that prevent AI workloads from reaching production.
The announcement, made at the Current London conference, comes as Confluent’s APAC leadership acknowledges a persistent gap: most AI projects in the region fail before deployment because the underlying data layer lacks the security and scalability needed for production-grade operations.
Key new capabilities
The updates span four areas. Natural language operations allow developers to use a managed Model Context Protocol (MCP) server as a control plane, letting AI build, manage, and debug streaming pipelines using natural language; Agent Skills add a layer that encodes organisational best practices for consistent execution. Automated data privacy introduces a built-in machine learning function for personally identifiable information (PII) detection and redaction directly in Flink SQL, without custom code or external services, unlocking AI use cases in regulated sectors including financial services and healthcare. Secure connectivity adds Azure Private Link support, keeping AI workloads off the public internet with private paths to Azure-hosted models and databases. Finally, unified engineering workflows bring Flink SQL into dbt, the industry-standard data pipeline framework, lowering the barrier to real-time adoption.
Confluent also now supports Anthropic and Fireworks AI models, as well as TimesFM for anomaly detection, directly within Flink stream processing workflows.
In APAC, the race to deploy AI is hitting a hard reality: most projects never make it past the pilot phase because the data layer is not secure or scalable enough for production. We are changing that, said Greg Taylor, Vice President and General Manager for APAC, Confluent.
The releases extend integration work announced at IBM Think 2026, where Confluent and IBM deepened connectivity between Confluent Cloud and the watsonx.data platform.
