Check the
documentation
Ask the
Community
Take a look
at
Academy
Cognite
Status
Page
Contact
Cognite Support
Dear community,
We are excited to share highlights from the Q3 2025 release of Cognite Data Fusion. This post will walk you through:
-
Atlas AI becoming smoother, smarter, and more governed
-
New tools for operators in the field to plan and act on checklists seamlessly
-
Richer context in Charts, 3D, and simulator integrations to connect data with real-world operations
And there’s much more to explore. Dive into the latest release notes and uncover all the features and improvements designed to accelerate your industrial data journey. These brand new additions will be available starting September 2nd. We’d love to hear your feedback here on Cognite Hub, and we thank all our community members for your contributions so far. Keep the ideas coming — your input helps us shape the future of Cognite Data Fusion.
Atlas AI
Smoother Agent Interactions
We’ve evolved the Atlas AI experience to feel like a natural extension of CDF. Cross-agent chat now enables users to carry conversations fluidly across agents, reducing friction and keeping work moving. A refreshed UI aligns visually with CDF for smoother, more consistent interactions. Together, these updates strengthen trust and make agents an integral part of daily workflows, driving higher engagement.
Smarter Knowledge Graph Queries
The new Agentic QKG v2 unlocks richer, engineer-like insights. By supporting multi-step queries, diagram annotations, and edge properties, agents can now handle nuanced industrial questions that previously ended in dead-ends. This makes it easier to retrieve deep context and uncover relationships across complex datasets.
Secure & Governed Agent Operations
Enterprise customers now have stronger governance. New access controls (agents:read, agents:write, agents:run) define who can view, create, or execute agents. High-impact tools like Python execution and REST API calls require explicit confirmation before running. This prevents accidental execution and strengthens compliance and security.
Accelerating Time-to-Value with Templates
Beta
We’re launching ready-to-use agent templates like Data Scout (for instant industrial data retrieval in Canvas) and Time Series Trender (for trend and root-cause analysis). These reduce time-to-insight and empower engineers to self-serve without waiting on data science teams.
Document Parser Enhancements
Beta
Parsing is smarter and more flexible. Users can leverage LLM vision capabilities, set custom parsing instructions in UI or API, select page ranges, and extract multiple instances. This unlocks more data trapped in files while giving customers more control and performance.
Industrial Tools
Checklist Scheduling
Beta
A new Schedules tab allows supervisors to plan upcoming checklists by shift, day, or week. This ensures the most critical work gets prioritized, giving supervisors flexibility to optimize operations and resource allocation.
Ad-Hoc Checklists from Mobile Templates
Not all fieldwork follows a schedule. With mobile checklist templates, operators can now generate unique, ad-hoc checklists directly in the field. Filtering options make it easy to find the right template, and mobile support means no dependency on desktop systems. This empowers operators to respond to changing conditions on the spot.
Activities Visualization in CDM Projects
Beta
Charts now allows engineers to overlay activities (like maintenance or process events) directly on top of time series. By clicking an activity, users can view full details instantly. This bridges the gap between “what happened” (data anomalies) and “why it happened” (operational context), speeding up troubleshooting and decision-making.
Smarter, Faster Ways to Explore Data
Search and browse just got a major upgrade. Users can:
- View denser tables with frozen headers.
- Resize preview panels for quick file inspection.
- Apply alphanumeric sorting, Enum filters, and “select all” options.
- See clearer tooltips and error handling.
The result is faster, more comfortable navigation, letting users spend less time finding data and more time analyzing it.
Documentation:
Introduced Space and External Id across filters, columns and properties
Filters/columns: “Select all” and filtering on enum properties
Improvements to the search configuration UI
Support for CDM and Hybrid Projects
We’ve closed a key migration gap: hybrid CAD projects and CDM point cloud contextualization are now fully supported via API. This unblocks major customer migrations and accelerates the transition from classic to CDM.
Automatic Tag Detection for 360° Images
Users can now run text detection on 360° images to automatically generate annotation suggestions with bounding boxes. This reduces manual contextualization time and encourages
more frequent updates to this rich data format.
New Measurement Tools
3D becomes more practical with enhanced measurement tools: horizontal cylinder measurement (ideal for exchangers and flanges), coordinate-based measurement, and improved UX for precision. These improvements reduce the need for on-site field work.
Power BI App 3D Viewer
Beta
Requested by customers like Aker, this Power BI app allows direct viewing, rotating, and measuring of CAD models inside dashboards. Users can now correlate Power BI data with 3D spatial context, providing a richer lens for analysis.
Extract Information Model from Simulators
Beta
A long-standing challenge has been the opacity of simulator files. This feature automatically extracts flowsheet hierarchies — equipment, properties, connections — into CDF. With initial support for GAP, engineers and AI systems can finally see and use this data without opening proprietary software.
External Dependencies in Model Revisions
Beta
Instead of uploading massive monolithic archives, users can now upload smaller ZIPs and manage dependencies as separate CDF files. This saves storage, avoids duplication, and makes updates much faster — particularly critical for large production networks.
Documentation:
Data Management
Configurable Tag Detection UI
Beta
The diagram parsing UI now offers a configuration interface. Users can fine-tune parameters (like minTokens and searchField) and scope which assets or files are searched. This delivers more accurate and relevant results without custom scripts.
Annotation Box Adjustments
Beta
When annotation boxes shift after file version changes, users can now move and scale them directly in the UI — even in bulk. This saves manual rework and keeps annotations aligned with minimal effort.
Core Data Model Support for DB Extractor
The DB Extractor now writes directly into CDM instance spaces, whether for Time Series, File concepts, or any other data model nodes and edges. This simplifies onboarding of database sources, shortens time-to-value, and makes CDM a first-class destination for structured and unstructured data.
Admin
App-Scoped Access
Admins can now scope access capabilities to specific applications. By limiting which groups apply for each app, queries in large Data Models are faster and more efficient. This change strengthens performance without reducing flexibility.
