The product has a name, a launch window, and a price: Meta will release “Hatch,” a consumer AI agent, within weeks, with a target of early September, according to a report from The Information on Aug. 24. The company also plans to introduce a new AI model, codenamed “Watermelon,” in October. The two launches represent Meta’s first serious attempt to sell AI directly to consumers, a shift from a company that has made its fortune from advertising.
Hatch is the product that will test whether Meta can do for AI what it did for social networking: build something millions of people use every day, and find a way to charge for it. The agent, according to the report, will be a paid consumer product, a departure from Meta’s history of free services supported by ads. The decision to charge reflects both the cost of running AI and the company’s ambition to build a second revenue stream.
The timing is pointed. Meta has been slower than its rivals to put consumer AI products in front of its users, and it has watched OpenAI, Google, and Anthropic build substantial consumer franchises. The September launch, if it holds, would put Meta in the market before the holiday season, with a product it can refine through the end of the year. The Watermelon model, arriving in October, would power the next phase of the product.
The strategy is unusual for Meta in one respect: the company is entering a market with an existing user base but no existing AI franchise. Its billions of users across Facebook, Instagram, and WhatsApp give it distribution that no competitor can match, but distribution alone has not been enough for AI products, which succeed on capability and trust. Hatch will have to be good enough to justify a subscription, and Watermelon will have to be good enough to power it.
The move also signals how Meta sees its future. The company’s core business, advertising, has been remarkably durable, but its growth depends on the attention economy, and the rise of AI assistants threatens that dependence: if users ask an agent for answers instead of scrolling a feed, the advertising model has a problem. Meta’s answer, according to people familiar with its thinking, is to own the agent, and to find a business model that does not depend on attention alone.
The competitive field is crowded and expensive. Consumer AI agents are being built by companies with far deeper AI research budgets than Meta’s peers, and the category has already seen a shakeout of products that failed to find users. Meta’s advantages are its scale and its data, and its challenge is converting both into a product that feels indispensable rather than incidental.
The report also raised questions Meta has not answered: the price of Hatch, the features it will include, and whether it will integrate with Meta’s existing products or stand alone. The company has declined to comment on the details, and people familiar with the plans say the launch date could slip, a common occurrence in AI product development.
For the industry, the launches are a test of a broader question: whether the technology companies that built the last era of the internet can build the next one. Meta built the largest social network in history by being first to understand how people connect; the AI era rewards different skills, and the company’s performance in it will be judged by Hatch’s reception.
The broader backdrop is the transformation of Meta’s business. The company has spent the past two years investing heavily in AI infrastructure, building the data centers and the model-training capabilities that its rivals have long possessed, and it has begun integrating AI into its advertising systems with results it describes as promising. The consumer agent represents the other side of that investment: a product that can generate revenue directly, rather than indirectly through better ads. The company’s leadership has said repeatedly that AI is its top priority, and the launches planned for September and October will be the most visible test of whether that priority translates into products people will buy.
The reaction from the market has been muted so far, reflecting both the crowded field and Meta’s history with new product categories. The company has launched and shut down consumer products before, and investors have learned to wait for evidence rather than announcements. But the stakes of Hatch are higher than the company’s earlier experiments: Meta is positioning it as a paid product, which means it must clear the bar of demonstrated value, and the Watermelon model must be good enough to power that value. The months ahead will show whether Meta can turn its distribution advantage into a subscription business, and whether the company that connected the world’s people can connect them to AI.
The next few weeks will provide the first evidence. If Hatch launches in early September as planned, it will enter a market that is simultaneously the most contested and the least settled in technology. Meta’s bet is that its distribution, its data, and its willingness to charge will give it a position that its rivals cannot easily copy. The product will be the argument, and the market will be the judge.
