One foundation for what’s next: classic, asset-centric services are retiring

Today October 15, 2026 we're announcing the deprecation of the classic asset-centric services in Cognite Data Fusion. They will be fully retired on December 31, 2027. The full list of affected services, and the details for each, is in our official deprecation notice.

Many of you have already made the move to Data Modeling and the Core Data Model (CDM). Since late 2024 we've recommended that every new project start on CDM, and all of our new product development has gone into Data Modeling. This announcement completes that shift. Running two parallel architectures slows us down. Focusing on one lets us put all our effort into the foundation that gives you the most value.

We know an upgrade takes real work, so we want to be clear about two things: why it's worth it, and what we've done to make it as smooth as possible.

Why Data Modeling

Describe your operations in your own language - The classic APIs gave every customer the same fixed shapes: assets, events, time series, files. Data Modeling lets you describe your operations in your terms. You can extend the Core Data Model with your own types, properties, and relationships, and adopt the industry standards that matter to your business. Your data then reflects how your organization actually thinks about its plants, equipment, and work.

Grounding AI - AI agents are only as good as the context they're given. A typed, richly connected knowledge graph gives Atlas AI agents and other AI tools much more to reason over. They can follow real relationships rather than guess at them, which means more accurate answers with more confidence behind them.

Build exactly what you need - Data Modeling is the foundation for Flows - Custom Apps. You can build and deploy applications on top of your own data model, using the data types you have today and adding new ones as your needs grow.

One foundation for everything new. New capabilities, from Records for high-volume event data to Data Products and Semantic Search, are built on Data Modeling. Upgrading means you get them all.

What's changing

There are two kinds of change:

• Full retirement. Some classic services will shut down completely and are replaced by Data Modeling equivalents. This includes things like Assets, Relationships, Labels, Annotations, Events, Sequences, and Data Sets APIs, plus Data Explorer and the pre-CDM versions of InField and Maintain. For example, assets become CogniteAsset instances and Relationships become Edges.
• Pattern shift. services stay available but move to a new pattern. These are Time Series, Files, and 3D. You'll no longer create them as standalone classic resources. Instead, they become part of your data model. Your time series datapoints and file content keep flowing through the same APIs. Only how metadata is managed and queried changes.

The deprecation notice maps each service to its replacement.

We've made upgrading as easy as we can

The migration plugin for the Cognite Toolkit gives you a guided, repeatable path from an asset-centric project to a data modeling-based one. It works in stages:

1. Prepare. Define your target data model and how your classic resources map to it, then deploy the migration model.
2. Migrate your data, one service at a time. Start with the foundations: data sets move to instance spaces, then source systems, assets, events (to CDM instances or to Records), files and time series, and annotations.
3. Migrate your applications. Bring across Canvas, Charts, InField configuration and data, 3D models, and 360 images, so your users keep their work.

You can also replicate production data into a development project and rehearse the whole migration before touching production. The guide lists known liimitations, so that you can plan around them in advance.

👉 Get started here: Using the migration plugin

Key dates

• October 15, 2026: Classic asset-centric services are deprecated. They keep working as they do today, but receive no new features.
• December 31, 2027: Classic services are retired and the APIs shut down.

What to do now

1. Take stock of your dependencies. Look beyond your CDF project to the integrations, dashboards (Power BI, Grafana), custom applications, and third-party or partner solutions that read from the classic APIs.
2. Build anything new on Data Modeling. Every new pipeline or use case built on the classic APIs adds to what you'll need to upgrade later.
3. Try the migration plugin on a development project with a copy of your data.
4. Talk to your Cognite team. Your account team can help you scope the upgrade, plan the timeline, and work through anything specific to your setup.

Have questions? Ask them in the comments below or reach out to your account representative.