Product Release Spotlight - October 2023 Release │ Cognite Data Fusion

We are excited to announce the October 2023 release of Cognite Data Fusion! This post covers some of the significant highlights of the release:

• Enhanced data exploration

• AI-powered data exploration

• Manual contextualization of 3D models

• Saving no-code calculations

• Monitoring asset health indicators

• Jupyter Notebooks with Cognite AI Copilot

These enhancements will be made available on October 31, 2023. We would love to hear your feedback here on Cognite Hub over the coming weeks. You can also find more detailed release notes on our Documentation portal, as well as the October Roadmap update on Product Updates. This month, we focus on data onboarding. 

We also want to thank all our community members for your contributions so far. We're eager to continue this collaborative process with you, seeking your valuable input on both existing and upcoming product features. Let's keep this momentum going!

Here’s a short video summarizing the exciting updates

Enhanced data exploration

We are happy to share our newest enhancement to the search experience within Cognite Data Fusion, furthering our ambitions to deliver easy access to complex industrial data. This feature provides user-friendly search and enhanced context for industrial users, such as engineers, technical experts, and operational teams. You can narrow your search to a specific site - like an individual offshore asset or a processing facility - to make the search results more relevant. The search results are grouped and presented in an industrial context and can be viewed through a list view and a 3D view with a wide set of filter capabilities across all given data types. You can easily change between layered models to see CAD models, 360 images, and point cloud models. 

AI-powered data exploration

As an advancement to traditional search and filtering of data models, users can now search using AI. You can toggle on AI and find data by using the example questions or write your own using natural language. You can at any time adjust the generated filters and also view the aggregates as charts and text summaries. 

Manual contextualization of 3D models

3D models such as CAD or laser scans don’t always contain the required information to link the model to assets, and current workflows require a fully contextualized 3D model to enable you to really benefit from the power of 3D. 

If you're adding 3D models to Cognite Data Fusion, you can now do manual contextualization of both CAD models and point cloud models in Cognite Data Fusion. The asset links can be reflected in all Cognite applications using the Reveal 3D viewer, enabling users to maximize value capture from applications such as InField and Maintain or any bespoke customer applications built with Cognite Reveal.

Saving no-code calculations

If you're using CDF to create and run calculations, you can now create, save, and schedule the calculations in Charts. This eliminates the need to create Python scripts and deploy functions. The scheduled calculations can be consumed on external dashboards or used as input to advanced models.

Monitoring asset health indicators

Monitoring asset health indicators and performance KPIs is one of the key jobs of engineers and subject matter experts when evaluating asset health. 

We’re now adding the ability to set up threshold-based monitoring in Charts, using raw time series or the saved no-code calculations. You can also enable email notifications to be sent out when thresholds are reached for the monitoring jobs. 

Jupyter Notebook with AI copilot

Data scientists and data engineers can now work with and run Python scripts from Cognite Data Fusion without setting up infrastructure. This is a valuable tool when combining the power of Cognite Data Fusion with external tooling, such as data analytics packages, and when you need to enrich your data using custom logic. Notebooks run in a sandboxed environment inside the browser under the current user credentials, ensuring data integrity.

We’re also introducing a code copilot integrated with Jupyter Notebook. By leveraging GenAI, the copilot will help you write and understand code. Code generated from natural language helps you get to value quicker by reducing the need to write boilerplate code or do advanced data analytics. 

We now support natural language data insights using large language models in notebooks. This allows you to ask questions about data retrieved from Cognite Data Fusion or to generate plots.

Early adopter programs

1. Data workflows is a new managed process orchestration service within Cognite Data Fusion. Using CDF Workflows, you can orchestrate the order and timely execution of interdependent processes such as transformations, functions, requests to CDF, and dynamic tasks. The service is only available by joining the Early Adopter Program. Request participation through the group on Cognite Hub.
2. Industrial data search enables industrial users without technical background to easily find relevant data. Search results show the industrial context with the ability to narrow down the search to a specific site/asset, and with several lenses/dimensions of searching and viewing data, such as via 3D, AI search, and standard search with list view and a range of filter capabilities. Please come and join us in the Early Adopter Program. Request participation through the group on Cognite Hub.