The Jarvis dream just won't die
Fourteen years after Siri, four labs are shipping the same proactive assistant within weeks of one another
In the days before Anthropic opened Code with Claude, its developer conference, in San Francisco this week, engineers were finishing a feature called Orbit. The hidden engineering toggle that gates it, according to TestingCatalog, a tech-leaks site, is named "tibro enabled" — orbit spelled backwards. Orbit is a proactive briefing system inside Claude Cowork that connects to Gmail, Slack, GitHub, Calendar, Drive, and Figma, and generates daily insights and suggested actions without being asked. It is also the same thing Meta, Google, and OpenAI are racing to ship in the same window.
Meta, according to The Information, is building a consumer agent called Hatch and a separate agentic shopping tool inside Instagram, the first targeting internal testing by the end of June and the second a launch before the fourth quarter. Google, according to Business Insider, is dogfooding an internal "24/7 personal agent" codenamed Remy, described in an internal document as a tool that "can take actions on your behalf — not just answer questions or generate content." OpenAI already has the cultural object — OpenClaw, the agent that became this year's tech-enthusiast obsession after Sam Altman, OpenAI's chief executive, hired its creator Peter Steinberger in February. The four largest American AI labs are now converging, within weeks of one another, on the same product category.
This article is for Vector members. Start a 7-day free trial to keep reading.
Start your free trial
