Meta is building an OpenClaw for people who live on Instagram

Meta is developing a consumer AI agent, codenamed Hatch, designed to shop on Instagram, manage inboxes, and operate across third-party apps. This ambitious project aims … Read more

Meta is building an OpenClaw for people who live on Instagram
Meta is developing a consumer AI agent, codenamed Hatch, designed to shop on Instagram, manage inboxes, and operate across third-party apps. This ambitious project aims to offer a user-friendly version of the viral OpenClaw agent. Meta is investing heavily in this personal superintelligence push, with a public launch anticipated by late 2026, despite significant trust concerns.

Meta is building a consumer AI agent that shops on Instagram, manages your inbox, and handles tasks across third-party apps, according to The Information. The project, internally codenamed Hatch, is Mark Zuckerberg’s attempt at a polished, mass-market version of OpenClaw—the open-source agent that went viral earlier this year.The Information reports Meta has been testing Hatch on simulated versions of DoorDash, Reddit, and Outlook, training it to work beyond Meta’s own properties. Inside Instagram, the plan is to let users buy products they spot in Reels—a direct shot at TikTok Shop. Meta’s recent move allowing creators to tag up to 30 products in a single video looks like quiet groundwork for that pivot.

Zuckerberg wants OpenClawminus the steep learning curve

On Meta’s earnings call last week, Zuckerberg called OpenClaw promising but too fiddly for most people to actually set up. His pitch is an agent that works on your goals around the clock—part of his broader personal superintelligence push. That same vision is why Meta just hiked its 2026 capex by $10 billion to roughly $145 billion, even as the company prepares to cut 10 percent of its workforce later this month.Meta tried to hire OpenClaw creator Peter Steinberger earlier this year. He picked OpenAI instead. Zuckerberg did, however, bring in the founders of Moltbook, the briefly viral AI agents forum.

Anthropic models are running Hatch for now, Muse Spark next

For now, The Information says Hatch is being tested on Anthropic models, though Meta plans to eventually shift it onto Muse Spark, its newer in-house model. A public launch isn’t expected before the end of 2026, with internal testing reportedly wrapping up by next month.A trust problem already shadows the launch. Meta reportedly wants users to plug in sensitive data—including health and financial information. One source described it to the Financial Times as “a trust deficit as wide as the Grand Canyon.” OpenClaw itself has had embarrassing public failures, most notably in February when Meta’s own director of safety and alignment, Summer Yue, watched it wipe her entire inbox as she frantically typed “STOP OPENCLAW.“CFO Susan Li also hinted Hatch could land on Ray-Ban Meta smart glasses, which she called “the best form factor for agentic interactions”—though even she admitted those capabilities are still in their infancy. Meta isn’t alone here either: Google is reportedly testing its own consumer agent, codenamed Remy, while OpenAI snapped up Steinberger. The agent race is suddenly crowded, but the harder question is whether mainstream users will trust any of these firms enough to hand over their inbox, calendar, and credit card.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *

About the Author

Easy WordPress Websites Builder: Versatile Demos for Blogs, News, eCommerce and More – One-Click Import, No Coding! 1000+ Ready-made Templates for Stunning Newspaper, Magazine, Blog, and Publishing Websites.

BlockSpare — News, Magazine and Blog Addons for (Gutenberg) Block Editor

Search the Archives

Access over the years of investigative journalism and breaking reports