Agentic AI, Connected Context, Faster Action: What’s New in Cumulocity Q4 2026
Industrial IoT is evolving beyond simply connecting and monitoring equipment. Increasingly, the opportunity lies in helping people understand operational data, add context and act on it more intelligently.
The Q4 2026 Cumulocity enhancements continue that evolution, with major advances in Agentic AI, digital twins, industrial data management and streaming analytics, alongside improvements that make dashboards and applications easier to build and use.
Let’s dive into what’s new.
Agentic AI
Cumulocity is evolving to provide the harness for industrial AI: connecting AI assistants and agents with live equipment data and enterprise context, while providing the controls needed to use AI safely in industrial environments. This direction brings together AI assistants, agent tooling, a semantic OT data layer and an AI Control Plane with policies, guardrails, human oversight and auditability.
AI Assistant – Public Preview
The Cumulocity AI Assistant enters Public Preview as a context-aware AI assistant designed to become the standard way to interact with AI across Cumulocity applications. It provides a common entry point for AI capabilities directly within the applications where operators and engineers already work.
AI Control Panel – Public Preview
The new AI Control Panel provides the foundation for governing AI within Cumulocity. Organisations can bring their own LLM provider, gain observability through OpenTelemetry and use an AI Circuit Breaker or “kill switch” when required.
Code Mode – Public Preview
Code Mode provides a secure sandbox where agents can execute code against Cumulocity APIs, giving them an efficient way to perform more sophisticated tasks while keeping execution within a controlled environment.
Looking Ahead: AI Acceleration Bundle
From January 2027, the AI Acceleration Bundle will bring these capabilities together with ready-to-use expert agents, tools for building your own agents, and an AI Control Plane providing policy enforcement, human-in-the-loop approvals, audit logging and the AI Circuit Breaker. A free preview of the AI capabilities is available today.
Cockpit
Q4 brings a range of Cockpit enhancements focused on making dashboards more powerful, flexible and easier to customise.
Enhanced Asset Table Widget
The Asset Table Widget has been rebuilt on the data grid, with advanced filtering, sorting and column customisation. It also adds a new Compact Mode to display more assets in smaller spaces, improved sorting for latest measurements and smoother navigation between devices and dashboards.
Enhanced AI SCADA Widget
The modernised AI SCADA Widget introduces AI-assisted SVG generation based on asset properties, measurements, alarms and events. Interactive placeholder mapping makes it easier to bind visualisations to live data, while faster rendering, CSS encapsulation and isolated scopes improve performance and security. Existing AngularJS-based SCADA widgets continue to work in legacy mode.
Custom Icons for your Assets
The improved icon selector now supports custom SVG icons across applications, managed objects and Digital Twin Manager assets. Icons can be defined once at asset-definition level, inherited by assets and overridden individually when needed—making it easier to create a consistent visual experience across Cumulocity.
Richer Data Graphs and Data Exploration
The Data Graph widget and Data Explorer now support multiple aggregation functions for the same data point, min/max bands and automatic aggregation based on the selected time range. Smart zooming provides more detail as users explore their data, while existing widget configurations continue to work without migration.
More Cockpit Improvements
Type dashboards can now be shared across applications, making common dashboard experiences easier to reuse. Angular 21 support is also available, with Angular 22 support planned for the Web SDK as Cumulocity continues to modernise its frontend technology stack.
Digital Twin Manager
Q4 brings operational data and asset context closer together, while simplifying how organisations work with large enterprise asset structures.
Data-on-Assets
Data-on-Assets connects device telemetry directly with Digital Twin assets. Data-point mappings automatically map measurements to assets, while additional context such as meaningful names, units and target ranges makes operational data easier to understand and use.
This creates a stronger foundation for reusable dashboard templates, default rules and KPIs, while providing richer context for AI-assisted applications and analytics.
Simplified Enterprise Asset Management
Several Asset API enhancements reduce the number of calls required when working with large asset hierarchies. Developers can create and assign subassets in a single request, resolve asset definitions for multiple IDs at once, For sub-assets assignment avoid unnecessarily returning full subasset collections, and retrieve parent references more efficiently. All changes are backward compatible.
More Languages for Digital Twin Manager
Digital Twin Manager now supports seven additional languages: Chinese, Dutch, French, Korean, Polish, Brazilian Portuguese and Spanish. Together with English and German, this brings DTM in line with the language options available across other Cumulocity applications.
DataHub 2.0
Industrial data architectures are changing. Open standards such as Apache Parquet and Apache Iceberg have become mainstream, while AI is creating new demand for scalable analytical access to operational data. DataHub 2.0 responds to these trends by moving from accessible IoT data toward agent-ready data.
Open IoT Lakehouse with Streaming Lake Ingestion – Public Preview
At the heart of DataHub 2.0 is an open IoT lakehouse with Streaming Lake Ingestion, designed around three principles:
- Simpler: Activate it and Cumulocity handles schema evolution and lake maintenance in the background.
- Fresher: Data becomes available within less than five minutes, supporting scalable long-term analytics and operational decision-making.
- Open: Apache Iceberg and Parquet provide integration with analytics platforms such as Databricks, Snowflake, AWS Athena and Microsoft Fabric.
This gives organisations an open foundation for bringing Cumulocity IoT data together with their broader enterprise data ecosystem.
Find more details here
Looking Ahead: Lake Analytics
The DataHub roadmap continues toward Lake Analytics, including time-series aggregation, AI Assistant integration through MCP and, longer term, agent-ready time-series analysis.
The goal is to prepare industrial data so AI agents can quickly work with trends, anomalies, KPIs and other analytical results—opening up use cases ranging from equipment insights and alarm triage to measurement correction and connectivity analysis.
Find more details here
Streaming Analytics
Q4 also makes Streaming Analytics and Analytics Builder easier to debug, troubleshoot and maintain.
Inline Debugging in Analytics Builder
New inline debugging brings an expandable log pane directly into the Analytics Builder canvas. Combined with the Logger block, users can “tap the wire” to inspect the values moving through a model, view logs relevant to the active model and make adjustments without leaving the editor.
The capability is initially available for standalone models, with support for templated models pending feedback.
More Streaming Analytics Improvements**
Full apama-ctrl microservice logs can now be accessed directly from the Streaming Analytics application, reducing the need to switch to Administration when troubleshooting.
Expression blocks also gain IF-THEN-ELSE support through the ternary conditional operator, making conditional logic easier to build within visual models.
Finally, the underlying Apama runtime is moving to Debian 13 and Java 25 for improved performance and security. Standard visual models require no changes, while customers using custom Java/EPL connectivity plug-ins or custom Python extensions should verify compatibility ahead of the rollout.
API Updates & Security
Q4 introduces further enhancements to Cumulocity’s security and connectivity capabilities, strengthening authentication, simplifying permission management and supporting more scalable device communication.
Enhanced Security
Cumulocity continues to strengthen its security posture with support for certificates without a Common Name, a centralised permission-checking service and certificate-based user authentication.
An important API change also affects password updates. Password changes through general user APIs are being removed, with applications and scripts required to use PUT /user/currentUser/password instead. On SaaS instances and 2027 yearly releases, the password field in general user API requests will be ignored.
And Much More
Q4 2026 continues Cumulocity’s evolution toward a more intelligent, open and context-aware approach to industrial IoT.
Agentic AI brings AI assistants and specialised agents closer to operational workflows. Digital Twin Manager connects live telemetry with richer asset context. DataHub 2.0 creates an open foundation for analytics and AI, while Cockpit and Streaming Analytics continue to make those capabilities easier to visualise, build and operate.
Together, these enhancements help organisations move more quickly from connected equipment and raw data to meaningful insight and action.