Meta is building an AI agent called Hatch and a separate AI shopping tool for Instagram. Both are expected before Q4 2026 as Zuckerberg pushes his AI investments toward real revenue.
Meta is building two AI products simultaneously that together suggest a clearer picture of where Mark Zuckerberg wants his company to go next. The first is an internal AI agent codenamed Hatch, designed to autonomously complete tasks on behalf of individual users. The second is an AI-powered shopping tool being built directly into Instagram. Both were reported by The Information on May 5, and both reflect the same underlying pressure: Meta has committed between $125 billion and $145 billion in capital expenditure for 2026 and needs its AI investments to start producing visible, revenue-generating products.
Hatch is being developed as a consumer-facing version of OpenClaw, an open-source AI agent framework. Meta is aiming to complete internal testing by the end of June, and to support that timeline, the company has built virtual environments that simulate real-world platforms including DoorDash, Etsy, and Reddit. Those sandboxed environments allow Hatch to be tested against realistic web interactions without putting live users at risk during the development phase. The team is also working on three specific capabilities that will define how useful Hatch feels in practice: its ability to decide when to act independently rather than waiting for instructions, how much information it can process at once, and a memory function that retains details across separate conversations. That last feature is particularly important for an agent designed to handle personal tasks, where context from previous interactions is what separates a genuinely helpful assistant from a system that resets every time you open it.
On the model side, Hatch is currently being trained on Anthropic's Claude models and is expected to transition to Meta's own Muse Spark model at launch. That sequencing is pragmatic, using best-available external models during development while preserving the commercial and strategic advantage of shipping on proprietary infrastructure.
The Instagram shopping tool is a separate effort targeting deployment before the fourth quarter of 2026. It builds on work Meta announced at the Shoptalk 2026 conference in March, where the company described testing a new experience that surfaces more product information and AI-generated review summaries when users click on ads or visit websites from within Facebook or Instagram. The new agentic shopping layer takes that further, moving from passive information surfacing to active assistance in the purchase process. The distinction matters because it represents a shift from Meta's platforms being places where users discover products to platforms where an AI agent can help complete transactions. That shift, if it works at scale, turns Instagram from an advertising channel into a commerce infrastructure.
The financial logic is straightforward. Meta's advertising business remains strong, but Zuckerberg has been explicit about wanting AI to open new revenue streams beyond ad placement. An agentic shopping tool embedded in Instagram sits at the intersection of two of the platform's most commercially valuable behaviors, browsing and purchasing, and gives Meta a way to extract value from the transaction itself rather than just the ad click that preceded it.
For the MENA region, both products carry direct relevance. Instagram is among the most actively used social platforms across the Gulf, Egypt, and broader MENA, and social commerce, buying products discovered through social platforms, is a fast-growing behavior in markets like Saudi Arabia and the UAE. Meta's AI shopping tool would land in an environment where the infrastructure for social-to-purchase conversion is still maturing, and where local and regional brands have been looking for better tools to close the gap between product discovery and completed sale. On the Hatch side, the Gulf's high smartphone penetration and growing appetite for AI-powered personal tools creates a natural early-adopter market, particularly among younger urban consumers in Dubai and Riyadh who are already comfortable with AI assistants embedded in daily life.
Meta's AI agent push is happening at a moment when nearly every major platform is racing to ship something agentic before the category solidifies around one or two dominant players. Hatch is not a product yet. It is an internal test with a June deadline. But the infrastructure being built around it, the sandboxed environments, the memory layer, the model transition plan, suggests Meta is treating this as a serious product cycle rather than a research project.
