Press note. Confluent today announced streaming agents, a new functionality of Confluent Cloud for Apache Flink that facilitates the Creation and scaling of AI agents who monitor, analyze and act on real -time data.
With this innovation, Streaming Agents eliminates barriers for artificial intelligence (AI) business level by unifying data processing and AI workflows, and provides easy and safe connections to all parts of a company, including large language models (LLM) and integration models, tools and other systems. Streaming Agents accelerates the adoption of agricultural AI, allowing more efficient workflows, a faster investment return time and the generation of new business models and opportunities.
«Agentic AI is part of the road map of all organizations. However, Most companies are stagnant in the prototype phasebeing lagging behind while others advance towards measurable results »explains Shaun Clowes, Product Director of Confluent. «Even the most intelligent AI agents act blindly if they do not have an updated business context. Streaming Agents simplifies the complex task of integrating the tools and data that create real intelligence, providing organizations with a solid basis to implement AI agents that promote significant change throughout the company ”he points out.
According to an IDC study, although organizations carried out an average of 23 generative concept tests between 2023 and 2024, only three reached the production phase. Of these, only 62% met expectations. Agents are as powerful as the tools and data they can access, but current workflows are extremely complex and expensive, which prevents companies from taking advantage of the entire value of agriculture. Although the existing AI frameworks facilitate the beginning with the agents, many teams have difficulty integrating data in real time in the initiatives of agriculture, which results in hallucinations and unreliable responses.
“Although most companies are investing in agricultural, their data architectures cannot support the autonomous decision -making capabilities that these systems require”Stewart Bond, Vice President of Intelligence Software and IDC data integration. “Organizations must give priority to the agricultural solutions that offer easy and safe integration and take advantage of real -time data to obtain the essential context necessary for intelligent action”.
Streaming Agents de Confluent
Streaming Agents carries the agricultural AI directly to streaming processes to help teams create, implement and coordinate events based on events with Apache Kafka and Apache Flink. Thanks to the unification of data processing and the reasoning of AI, agents get access to updated contextual data from real -time sources to adapt quickly and communicate with other agents and systems as the conditions change. Streaming Agents are always active and work on behalf of the companyoperating dynamically, processing high volume data flows and instantly responding to real -time signals with a context sensitive reasoning, as human operators would.
For example, Streaming Agents can carry out a competitive assessment continuously supervising prices in electronic commerce sites and automatically updating products on the website of a retailer to offer the most competitive offer to customers. The main streaming agents characteristics are:
- Call to tools for context automation: The invocation of tools through the model context protocol (MCP) allows agents to select the appropriate external tool, such as a database, software as a service (SAAS) or API, to carry out significant actions. The call call takes into account what is happening in the business and what other systems and agents are doing.
- Connections for safe integrations: Surely connect models, vector databases and MCP directly by Flink. The connections also protect the confidential credentials, promote greater reuse by sharing connections between multiple tables, models and functions, and centralize management for large -scale implementations.
- External tables and search to increase the accuracy of AI: Make sure the streaming data is enriched with data sources that are not from Kafka, such as relational databases and API Rest, to provide the most up -to -date and complete vision of the data. This improves the accuracy of the decision -making, the vector search and the increased recovery generation (RAG) applications, reduces the cost and complexity through the use of Flink SQL and takes advantage of the confluent cloud safety and network capabilities.
- Reproduction capacity for iteration and safety: Agents can be developed and evaluated using real data without real -time side effects, allowing hidden releases, A/B and faster iterations tests.
Streaming Agents is now available in open preliminary version.
