We’re excited to roll out major updates to the GoldenGate platform including: Oracle GoldenGate 26ai (for all major platforms, Oracle, Non-Oracle, Big Data etc.), GoldenGate Stream Analytics 26ai and Veridata 26c available for on-premises and OCI. This release brings a whole new AI Service for GoldenGate deployments, automatic schema evolution, AI Database 26ai support, greatly expanded source/target coverage and an upgraded compare-and-repair experience to simplify enterprise data movement and improve data integrity at scale.

What’s new in GoldenGate 26ai

Embedded AI service

  • GoldenGate 26ai introduces a new AI Microservice that brings AI capabilities directly into GoldenGate deployments. This service enables embedding generation for GoldenGate targets, supports bring-your-own models (local or remote), and provides a foundation for enriching real-time data flows with AI-driven context
  • The AI Microservice establishes the platform for future innovations such as real-time named-entity recognition, PII identification on transactional data, natural-language administration, agentic APIs (such as MCP), data enrichment using any LLM service, automated data quality enhancements, and intelligent auto-tuning based on real-world workloads. Together, these capabilities position GoldenGate as the best way to align real-time data movement with AI-native data platforms

Replicat Mapping Example:

MAP source.table_a, TARGET target.table_b,  
colmap (usedefaults, 
vector = @AISERVICE(embed,’cohere.embed-English-v3.0’,d)

Automatic Schema Evolution

This capability enables Oracle GoldenGate to automatically detect and propagate supported schema changes as part of the replication flow, reducing manual intervention during schema evolution scenarios.

  • Replicate table structure changes across Oracle, MySQL, PostgreSQL, DB2, SQL Server, and Snowflake
  • Options to automatically create or modify target tables during initial loads and online apply is provided
  • Schema evolution filtering utilizes the familiar INCLUDE/EXCLUDE syntax from the DDL parameter, and include options to override default data type mapping.
  • Troubleshooting and error handling is made easy through the use of existing DDL options
  • For Oracle-to-Oracle environments, native DDL replication remains the recommended approach (to cover the widest possible range of DDL statements)

Platform, Usability and New Supported Technologies

  • PostgreSQL capture now supports the PGOUTPUT plugin as an alternative to TEST_DECODING, enabling native logical replication, improved performance, and better alignment with PostgreSQL best practices
  • GoldenGate 26ai Air-Gap Data Streams extend real-time data movement into disconnected and highly regulated environments, enabling trusted data sharing where connectivity is limited or tightly controlled
  • COLMAP enhancements automatically add mapped columns to the target tables (AutoSchema), enabling simplified schema evolution, reduced manual DDL management, and faster onboarding of generated columns
  • Unified web UI changes to present a single unified UI to manage all processes in a deployment, users may now access to Distribution / Receiver Services, Performance, Extracts, Replicats all in a single web page
  • Enhanced Numeric Processing, full heterogenous support has been added for special numbers, including positive and negative Infinity, not a number, and signed not a number
    • support has been added to @COLTEST and @COLSTAT for these special values.
  • Oracle AI database 26ai capture and delivery support
  • Support for Capture and Delivery of YugabyteDB and YugabyteDB Anywhere
    • With Oracle GoldenGate for Yugabyte, you can perform capture and delivery from PostgreSQL source to Yugabyte target database on a Linux platform
  • Support for Capture and Delivery for EnterpriseDB Postgres Advanced Server
  • Delivery to Google Cloud Spanner

Security and credentials

  • GoldenGate 26ai introduces a new plug-in framework that enables integration with externally managed credential stores. The framework provides a flexible API for third-party integrations, works across all GoldenGate source and target data stores, and supports multiple languages including Java, Go, C/C++, and Rust

What’s new in Veridata 26c

  • New features
    • Built-in job scheduler and server configuration management via the UI.
    • Compare and repair support for MongoDB (including Atlas, AJD, JCT), Amazon DocumentDB, Azure Cosmos DB for MongoDB, MariaDB, and SingleStore.
    • OCR image
  • Deployment
    • Available as a managed OCI service for streamlined setup and operations and for on-premises customer managed environments.

Stream Analytics 26ai – key new features

  • Included with the OCI GoldenGate platform, enables real-time stream processing and analytics supporting data enrichment, transformation, and quality checks, along with time-series, geospatial, and AI/ML use cases
  • New features include:
    • Updated user interface adopts same Oracle Redwood themes for both Cloud-native as well as for on-premises deployments
    • Read data events directly from GoldenGate Data Streams, eliminating the need for costly and complex Kafka clusters when your primary source of real-time data is CDC events from databases
    • Tight integration with Oracle AI Database 26ai with AutoML functionality, directly leverage in-database machine learning from your streaming pipelines
    • Run AI prompts directly from your streaming pipelines by linking out to LLMs, this brings many AI use cases into the streaming realm, such as PII identification, text summarization, NER, data enrichment and data cleansing operations on the data
    • Generate vectors from your streaming data pipelines and load the downstream database targets

Architecture, HA, and Multicloud

  • Prebuilt GoldenGate containers on Oracle Container Registry provide a greatly simplified way to always get the latest complete bundle patches and help you better utilize GoldenGate in your own Kubernetes and OpenShift ecosystems
  • Ongoing expansion for OCI GoldenGate in Oracle Database@Azure with more than a dozen regions available today and many more planned over the coming months, and OCI GoldenGate will launch as part of Oracle Database @ GCP in March
  • Flexible architecture patterns support a range of deployment models including source-side, target-side hub, and multi-hub topologies across on-premises, OCI, Azure and GCP(Coming),  Kubernetes-based environments, and self-managed deployments in third-party cloud platforms
  • Stay tuned: later in the GoldenGate 26ai product lifecycle you will see a brand new built-in GoldenGate high availability that removes shared filesystems and clusterware requirements allowing for easy deployment in on-premises environments or self-managed in 3rd party cloud platforms (like Azure, GCP and AWS)

Oracle University Advanced Courses

  • Oracle University is launching an advanced course for Oracle GoldenGate. This covers HA/DR, active-active replication, transformation, and there are lessons on the Rest API, and troubleshooting
  • The Advanced curriculum is a supplement to the already available GoldenGate Fundamentals class and the GoldenGate Certified Associate exams

Upgrades, Release Cadence and Support

  • GoldenGate 26ai is a release update (23.26.x) delivered as a bundle patch for existing GoldenGate 23ai customers. Please refer to the release notes for guidance on behavior changes, deprecated functionality, and desupported features
  • Customers on GoldenGate 19c and 21c may upgrade directly to GoldenGate 26ai and customers on “classic architecture” should consider using the provided Migration Utility located on My Oracle Support (MOS) portal
  • GoldenGate 26ai is considered a “long term support” release and will be eligible for Premier Support through at least 2031

GoldenGate Product Links:

This was a long blog post but really this is just the start of a great story for your GoldenGate 26ai journey. For more details, please take a look at:

Conclusion

GoldenGate 26ai is more than a feature release it marks a clear inflection point in how real-time data movement, streaming, and verification come together in an AI-native world. With embedded AI services, automatic schema evolution, expanded platform support, stronger security foundations, and a unified operational experience, GoldenGate continues to evolve from a best-in-class CDC engine into a complete real-time data platform.

Across GoldenGate 26ai, Stream Analytics 26ai, and Veridata 26c, the focus is consistent: simplify operations, reduce friction at scale, and give customers the flexibility to move, enrich, analyze, and trust their data wherever it lives and however it’s consumed.

And this is only the beginning. Over the GoldenGate 26ai lifecycle, you’ll see continued innovation across high availability, AI driven automation, streaming, security, and platform integrations.

We’re excited to see how you use these capabilities and where your GoldenGate 26ai journey takes you next. Stay tuned and welcome to the next chapter of real-time data, built for the AI era.