Like many of its peers and competitors, industrial software provider Cognite talks a lot about AI. But unlike some of those competitors, Cognite has real customer deployments — at scale — and used its recent Impact event in Houston to showcase:

  • NOVA Chemicals. Described as the largest producer of polyethylene in Canada, NOVA Chemicals started using Cognite’s data platform, Cognite Data Fusion (CDF), in one plant and now aims to have it running in 11 by the end of the year. CDF reduces data discovery time by up to 30% within individual sites, and the next phase of the project will see NOVA Chemicals move toward a single, enterprisewide knowledge graph.
  • TotalEnergies. TotalEnergies, a French integrated energy company, is deploying CDF across 39 assets over the next three years, combining data from over 200 key production systems. The initial focus is on health and safety, especially for frontline workers, but the project team has ambitious plans to streamline data acquisition and drive a range of AI-enabled data products and services across the business.
  • AkerBP. AkerBP operates six major assets off the Norwegian coast. AkerBP and Cognite share the same parent, and the two companies have worked together for years. AkerBP has bold digital plans for its newest asset — Yggdrasil — an oil exploration site intended to be remote-first, future proof, and capable of being operated most of the time by just 2 onshore staff. CDF underpins efforts to map, harmonize, manage, and use data from key systems throughout the design, commissioning, and operations workflows. Even early in the commissioning for Yggdrasil, AkerBP is using agents built on top of Cognite’s AI tools to automate extraction and normalization of data from suppliers’ PDF files and to cut root-cause analysis workflows from weeks (and sometimes as long as 9 months) down to hours.
  • Cosmo Energy. Cosmo Energy, a petrochemical firm, has deployed CDF across all three of its Japanese refineries, storing 250 billion data points and 5 million engineering documents about over 1 million pieces of equipment in those plants. Shifting from reactive to predictive maintenance and accelerating digital skills transfer within its workforce are the big focus areas for Cosmo Energy’s work with Cognite: AI agents support engineers on site to identify and resolve problems quickly. Behind the scenes, the Cosmo Energy team is also doing some impressive work to calculate the benefits of their investment in Cognite.

CDF Is The Foundation

Cognite has always talked about CDF. It helps customers combine data from different source systems into an industrial knowledge graph, ensuring consistency in how information is stored, structured, and described. CDF was an early example of the data aggregation capabilities that are so important to Forrester’s concept of the digital industrial platform, which we’re revisiting in new research that will be published soon.

Cognite leaders may claim to manage “the world’s largest industrial knowledge graph,” but they also recognize that Cognite can’t do this alone. At a pragmatic level, an AkerBP speaker spoke to us about how they divide the company’s cloud-based data in Microsoft’s Azure Cloud: operational data lives in CDF (on Azure), IT and business intelligence data lives in Microsoft Fabric (on Azure, of course), and subsurface prospecting data lives in the Azure Data Manager for Energy. Cognite’s partnership announcements with Databricks and Snowflake focus on making it easier for shared customers to manage data across these systems, avoiding unnecessary copying and duplication.

Atlas Makes This Useful At Enterprise Scale

Knowledge graphs are great, powerful, and an idea I’ve kept an eye on since I worked at a semantic web company 20 years ago. It’s also unfortunately true that graph platforms aren’t necessarily the most straightforward way for normal people to interact with huge volumes of complex and interconnected data. Data scientists? Yes. Ontologists, knowledge managers, and others who care deeply about these things? Yes. Regular people who just want to know which pump to service before they can stop for lunch? Less so. Cognite products like InField offer those frontline workers easier access to the insights they need, powered by an industrial knowledge graph they never need to think about.

With Atlas AI (not to be confused with OpenAI’s new AI browser of the same name), Cognite offers developers and data scientists an easier way to tap into and exploit the capabilities of CDF’s knowledge graph. There are the inevitable AI chatbots to the right of (almost?) every application window, with which any authorized user can have a conversation to query the graph. But there’s also a rich and increasingly capable AI agent workbench to support the creation of new agents, and even agentic workflows: For example, AkerBP described several use cases in which groups of agents were designed to collaborate on specific problems like root-cause analysis investigations.

Long-Term Solutions Need Good Foundations — And Something On Top

Far too many firms get overexcited by the hype and want AI-powered miracles right now, bypassing (or ignoring) hard work. If they’re lucky, they might produce an intriguing proof of concept. If they stay lucky, it won’t fail too catastrophically as they try to move it into production. Cognite isn’t really for those corner-cutting, risk-taking firms, and neither are its direct competitors. Cognite’s products (and those of its competitors) are for the firms that know this is slow, hard, and expensive — but ultimately rewarding. They know that someone has to think about data context, lineage, provenance, semantics, and all the rest of it. They know that the shiny AI app that pleases the board or impresses the prospective customer is a tiny little piece of the iceberg, resting atop a mountain of data operations groundwork.

If you’re a Forrester client and would like to explore this further, reach out to schedule an inquiry or guidance session.