The emergence of AI using vector databases, large language models, machine learning predictions, and AI Agents employing activity workflows is revolutionizing the way we can process and analyze steaming data. Instead of manually configured logic with conditions and rules, we can now employ intelligent decisions by AI models. We are happy to announce GoldenGate Stream Analytics 26ai with a comprehensive set of AI capabilities.
GoldenGate Stream Analytics (GGSA) provides these foundation capabilities for stream processing and time-series analytics. Stream Analytics works natively with GoldenGate’s replication capabilities and empowers customers to take data events from the data layer and process those events seamlessly within exact-once, ordered and transactionally-consistent streaming pipelines.
The next generation Stream Analytics 26ai capabilities include a host of new functionality designed to make it simpler and more intuitive to use AI & ML within streaming pipelines, here’s a summary:
Data Streams Integration
Stream Analytics 26ai now provides a direct integration with the GoldenGate 26ai Data Stream feature. Data Streams is an AsyncAPI-based interface to consume change events; pipelines can now consume GoldenGate events directly without a target-side GoldenGate instance or an intermediate Kafka topic. Users can either create new data streams from the GGSA console or connect to existing data streams. Pipelines can either opt for a single-table or multi-table input; it is now possible to process multiple table shapes within the same pipeline, making it easy to implement data fabric architectures.
Modern User Interface
The user interface has been refreshed with the Oracle Redwood user interaction design; we are using the latest UI frameworks for a beautiful low-code user experience.
AI Agent Pattern with LLM/GPT Processing
A new AI Agent Pattern allows users to utilize the OCI Generative AI Agent service to process events without any coding. The AI Agents use large language models (LLMs), retrieval-augmented generation (RAG), and custom tools to process incoming data based on natural-language prompts. For example, an agent can introspect event content and make intelligent decisions based on factors contained within the event or available through additional documents and database tables.
Create and Query Vector Embeddings
GGSA 26ai makes it easy to both create vector data and use it to perform AI operations such as similarity searches. The database target in pipelines has been enhanced to write to vector columns in Oracle AI Database 26ai and calculate embeddings based on ONNX models. In a Similarity Search pattern, you can find matches based on vector distance. For example, a pipeline can find a similar support case in a support database by comparing a new problem description with existing ones.
AutoML with Oracle AI Database 26ai
GGSA integrates with Oracle Machine Learning (OML) and AutoML to perform predictions and scoring of event data. Business users without an extensive data science background can use AutoML to create and deploy machine learning models, and GGSA can seamlessly integrate these models to identify trends and predict future behavior. This can be used in various use cases like fraud detection, predictive maintenance, and forecast models.
With these new capabilities in real-time AI and ML capabilities and an easy-to-use low-code editor, it is easier and more powerful than ever to automate stream event processing and get value from your real-time data!
Please visit us at Oracle AI World 2025 to see the new GoldenGate Stream Analytics 26ai as well as many new capabilities in GoldenGate and other Oracle products in action! Visit our session Streamlining Real-Time Analytics with GoldenGate 23ai to learn more about our product and customers.
Key GoldenGate links: