Cognite Product Roadmap - April 2023 Update

Coming into spring of 2023, we are happy to share an updated perspective on the roadmap for Cognite Data Fusion, and some key improvements to look forward to for the remainder of 2023 and into early 2024.

We’re continuing to see great engagement from you in Cognite Hub, your ideas and contributions are invaluable in deciding what to prioritize and improve now and going forward. Since January, we have averaged over 3,000 active users on Cognite Hub per month, and have seen a total of more than 1,600 posts from you. More than 30% of your submitted ideas have entered our backlog or have already been implemented. Thank you for your contributions and support.

For an update on the progress of our application suite focusing on Asset Performance Management (APM) use cases, please read this recent press release highlighting some of the key improvements.

We have improved Cognite Data Fusion in a lot of different areas over the past six months, but some key areas worth highlighting are:

1. Ship the first iteration of our Data Modeling capabilities to general availability - enabling CDF to speak your domain language, and unlocking a lot of new use cases and a next level of simplicity in working with data in CDF.

2. A wide range of improvements around data exploration in Cognite Data Fusion - global search, better usability in general, richer 3D capabilities (including support for point clouds and 360 images) and an improved document search and exploration experience.

3. A substantial improvement to our connectivity catalog, moving from a few select data extractors and a generic toolkit, towards a rich connectivity catalog with easier setup and management experiences both of extraction and data transformations.

As with all parts of the product, none of these areas are considered “done” - as we ship improvements we also unlock new ideas and opportunities to further improve the experience and capabilities of the product. As usual, your feedback is important to us when considering what the next steps should be.

Cognite and the Generative AI Revolution

Before we get into the details of the roadmap, we should call out the most important current trend, and one that is obviously very relevant to industry and data - what is Cognite’s thinking on the AI revolution and the rapidly evolving space of Generative AI and Large Language Models (LLMs)?

First of all, we think everything that is happening is extremely exciting, and we might finally be on the brink of unlocking the true potential in the vast amounts of data that we know are the key to solving the industry problems that were at the root of Cognite’s inception. We have already started publishing our thoughts on this revolution, watch our blogs and newsletters for more content going forward. In addition to this, our engineering teams are already actively working on several product features leveraging Generative AI, and as of April 25th, we have made the first one available to customers (challenge: see if you can find it). As we’re looking towards the rest of the year, we think Generative AI will unlock new opportunities in Cognite Data Fusion at several levels:

1. As an enabler for new and significantly improved user experiences. Leveraging LLMs, combined with the deep knowledge graph in Cognite Data Fusion for co-pilots, assistants and sometimes even a fully chat-based interface to your data holds great productivity potential. Expect to see several offerings in this space in the coming months.

2. As an enabler for richer contextualization, combined with the greater expressive power in our data modeling capabilities, generative AI can be a game changer. This applies not only to textual data but also to our ability to contextualize imagery and point clouds.

3. AI will embed itself alongside the existing sources of data and add additional data streams. Analyzing imagery, video and text feeds, AI will be able to generate higher-value events like anomaly detection or object detection. Predictive models and physics simulators are already a known source of input data, but the power of generative AI will be able to supplement these.

Illustration of ongoing experiments on industrial copilot.

With that topic covered, let’s move on to other parts of the product and talk about some of the more important developments planned.

Data Onboarding

In the area of Data Onboarding, we have three main areas of improvement:

1. Continue shipping more out-of-the box connectors to commonly seen systems and protocols. In June you can look forward to our first version of an SAP connector (also supporting writeback). We’re also investing in making the setup and running of extractors easier, as well as considering how we can open up the connector ecosystem to external partners and contributors in a better way.

2. Improve the user experience of setting up data onboarding - both extractor setup and management, as well as the data transformations. This also includes moving away from schedule-driven data transformations to a more event-driven and flow-based approach.

3. Harvest the experience from users of the new Data Modeling capabilities in Cognite Data Fusion and figure out how the tasks of data onboarding (including contextualization jobs) should fit best with the use of a richer set of data models.

Contextualization and Digital Twins

For contextualization and enriching data in Cognite Data Fusion, the new opportunities with Data Modeling and AI are an active area of focus, as we have already discussed. You can expect improvements to both contextualizing documents, improved understanding of images and point clouds (identifying equipment and other assets, etc.) as well as horizontal improvements related to usability and automation driven by generative AI approaches.

On the persistence side, the key focus for the coming period will be to continue strengthening our Data Modeling capabilities. We will be providing richer capabilities to interface with data (GraphQL mutations, potentially other protocols like OData and SQL), as well as extending the performance envelope of the underlying services. As a guiding statement, we aim to extend the supported performance envelope of the data modeling services from current levels of 5 million nodes and edges to upwards of 50 million  towards the end of 2023. In addition to this, we will be shipping change data notifications both for time series data and data in Data Models, enabling end-to-end event-driven architectures to be built around Cognite Data Fusion.

End User Tooling and Developer Experience

For our end users, and especially focusing on consumers of data, we will continue the improvements of our data exploration capabilities. Incorporating Data Models fully in data exploration, we are also looking at how we can make data exploration more accessible and attractive to business users - key topics are usability, focusing on relevant parts of the digital twin (site, line, etc.) and incorporating and improving 3D data exploration better.

In addition to improving the already existing end-user tools in CDF around exploration, working with time series and alerting, we will also be shipping the initial versions of what we think of as a “digital workbench” for working with all types of industrial data (time series, tabular data, images, 3D and more) to get an overview and solve a wide range of problems. Dubbed the “Industry Canvas”, we hope to launch the first iteration of this concept around summer, and we look forward to sharing our thoughts and getting your feedback.

Illustration of Industry Canvas concept under development

As for the developer experience on top of CDF, you can look forward to us building on the enabler that is Data Modeling, and leveraging that for a wide range of improvements. Look for extensions to the GraphQL endpoints (most notably support for mutations), developer-facing copilots to increase productivity, auto-generated SDKs, embedded jupyter notebooks for a quick way to get started, and much more.

Jupyter Notebooks embedded in Cognite Data Fusion.
 

We look forward to sharing more on these capabilities as they mature and are released. If you want to follow our development and shaping of these and many other features of Cognite Data Fusion, use Cognite Hub to engage with us and stay up to date with the latest developments.

Cognite Roadmaps are forward-looking and subject to change.