New marketing AI products

A lot of “artificial intelligence for marketing” still means a chatbot sitting on top of a spreadsheet. This week’s product moves were different: Cursor pushed deeper into tools that can actually build and test marketing dashboards, while Julius pushed deeper into tools that can actually analyze campaign data and customer groups. (cursor.com) (julius.ai) Cursor is best known as an artificial intelligence coding editor, not a marketing app. Its new pitch is that an “agent” can now work across repositories, show screenshots of what it changed, and hand work back to a human inside a rebuilt Cursor 3 interface. (cursor.com 1) (cursor.com 2) That matters because a marketing dashboard is usually built by front-end engineers before a marketer ever sees a chart. Someone has to wire up the page, connect the data, and check that the same screen works on a laptop browser, an Apple iPhone simulator, and a Google Android simulator. (cursor.com 1) (cursor.com 2) Cursor had already moved in that direction with its browser tool, which reached general availability in Cursor 2.0 after an earlier beta. The company says the browser can be embedded in the editor and can pass page information from the document object model — the structured map of a webpage — back to the agent. (cursor.com) Cursor also spent the past few months turning the agent into something that works away from the desktop. Its web and mobile rollout let people start or monitor agent work from the browser, Slack, GitHub, Linear, and mobile, then review the results later on a laptop. (cursor.com 1) (cursor.com 2) Then Cursor bought Graphite in December 2025. Graphite built code review software used by “hundreds of thousands of engineers,” and Cursor said the bottleneck in production software was no longer writing code but reviewing changes, merging safely, and collaborating without breaking things. (cursor.com) Put together, that gives Cursor a lane inside marketing work that does not look like marketing at first glance. If a growth team wants a new attribution dashboard or a campaign reporting page, Cursor is trying to own the engineering layer where that interface gets built, tested, reviewed, and shipped. (cursor.com) (cursor.com) Julius is coming from the other side. Its product is not for writing the dashboard code; it is for asking questions of the data once the spreadsheet, comma-separated values file, or warehouse export already exists. (julius.ai) (julius.ai) The company’s marketing pages now put cohort analysis near the center of that pitch. Cohort analysis means grouping customers by a shared starting point — like the month they signed up or the campaign that acquired them — and then tracking how each group behaves over time instead of averaging everyone together. (julius.ai) (julius.ai) Julius is also leaning hard into return on investment questions that sit one layer above raw reporting. Its marketing materials promise trend spotting, return on investment tracking, and connections to “spreadsheets to data warehouses,” which is exactly the unglamorous data plumbing that slows down campaign analysis in most teams. (julius.ai) (julius.ai) The interesting shift is that these two launches are not competing for the exact same user. Cursor is moving toward the engineer who has to make the reporting product real, while Julius is moving toward the marketer or analyst who has to explain why one customer group kept buying and another one disappeared after a paid campaign. (cursor.com) (julius.ai) That is a more mature shape for artificial intelligence software than the broad “ask anything” wave of 2023 and 2024. One company is specializing in the build-and-ship layer, the other in the analyze-and-decide layer, and both are aiming at the same budget line: marketing teams that are tired of waiting on custom dashboards and tired of exporting the same spreadsheet every Monday. (cursor.com) (julius.ai)