|
Microsoft, Databricks, SAP and other AI software providers are fighting over control and access to enterprise data known as the semantic layer, as we reported last week. Semantic layers are essentially a data management tool that companies use to standardize definitions for their business metrics, such as revenue and customers, and are now enabling AI to work more accurately and efficiently inside enterprises. Retailer Lowe’s, for instance, recently began using its semantic layer and another tool that defines how different types of corporate data are connected to each other—knowledge graphs—to improve how its AI agents help customers with orders or helping store managers coordinate the day-to-day work that employees do in stores, said Chandhu Nair, a senior vice president. Lowe’s runs an AI-powered shopping assistant for customers and an AI sales coach for its employees, both of which it developed with OpenAI over the past couple of years. Lowe’s also developed a customized agent that helps its finance teams verify that invoices are processed accurately, which is important because of the volume of invoices it handles as the fifth-largest importer in the U.S., Nair said. Tapping the semantic data and knowledge graphs became possible due to open-source tools the company used to develop its internal AI platform, which it uses for everything from developing agents to tracking employees’ token consumption. Advanced reasoning models and agent-development tools from OpenAI, Anthropic and Google, which are Lowe's main AI providers, also improved the capabilities of the platform, said Nair. If a Lowe’s AI agent detects that a customer’s order is delayed, he said, it can work with other systems that track order information, inventory levels and data about product availability from suppliers to identify a more precise cause of the delay and suggest ways to fix the problem, Nair said. (If that sounds similar, it’s because a sensor component firm in this story last week uses AI for a similar purpose involving its supply chain.) Lowe‘s also uses multiple agents to coordinate and maintain product inventory levels during times of heavy demand, such as promotions, which isn’t possible without the contextual data it is now able to use, said Nair. “If the merchandising team, the promotions team, and the inventory team have the same context about the decisions they make, they can make much more effective decisions,” he said. “That is a critical part of how we run our business today, and that context layer is a differentiating factor.” Another potential benefit of contextual data is that it could make agents cheaper to run, as they don’t have to “think” as much to make decisions, though Nair said it’s too early to tell if Lowe’s AI costs are dropping after giving its agents access to this kind of data. But there’s no doubt that giving AI better data to work with will be necessary to convince large companies to expand AI budgets at a time when many customers, including sophisticated ones like Uber, are still struggling to measure a return on their AI investments. Salesforce’s Subscription Revenue Split and What it Could Mean Salesforce did a curious thing in its first earnings report for its fiscal 2027, splitting its subscription revenue products into two buckets. One of them contains older applications like Slack, customer management and marketing automation while the other includes newer products like Data 360 (formerly known as Data Cloud) and Headless 360, a recently unveiled set of tools that lets customers use AI agents from companies like Anthropic to access their Salesforce data. The two new buckets show Salesforce is trying to transition from selling traditional business applications to products that customers can use to develop and use agents that, for example, process invoices and resolve customer service inquiries. Revenue for Salesforce’s traditional applications grew around 9% in its April quarter compared to last year, while its AI products grew around 25% during that period. (The newer AI products haven’t boosted its overall sales growth much, however. Read more here.) Vernon Keenan, a software industry analyst who covers Salesforce, sees the move as a way to highlight the growth of its AI products. “By isolating seat-based revenue into a new bucket, Salesforce is making a slight nod to the SaaSpocalypse by making sure growth in agentic consumption won’t get diluted by the expected decrease in the number of seats sold over time,” he said.
|
Gift Subscription
