Meta is preparing to launch a consumer AI agent called Hatch in the coming weeks and release a new model known as Watermelon in October, according to reporting based on internal documents. The plans would give the company a new route into paid AI products as it looks to turn substantial investment in artificial intelligence into revenue beyond advertising.

The reported Hatch launch is especially significant because it appears aimed at completing tasks across third-party services rather than simply answering questions in a chat window. The agent is expected to connect with services including DoorDash, Etsy, Reddit, Yelp and Outlook, while presenting users with a customizable dashboard that could include tools such as a fitness tracker or trip planner.

The details have not been confirmed by Meta in the supplied evidence. The Decoder attributed the plans to internal documents seen by The Information, while a separate TradingView item reported the same broad development but provided no additional article text.

What Meta is preparing

Hatch is described as a consumer version of OpenClaw, an existing Meta AI agent project. The distinction matters: a consumer-facing agent would place Meta closer to the emerging market for software that can take actions on a user’s behalf, including finding products, organizing information or managing routine tasks across online services.

The reported integrations suggest that Meta is considering Hatch as a personal operating layer for everyday activities. A user might access services such as DoorDash or Etsy through a single interface, while Reddit, Yelp and Outlook could provide information or support planning tasks. The evidence does not establish how much of this functionality will be available at launch, whether the integrations will rely on formal partnerships, or what permissions users would need to grant.

Meta is also reportedly developing a platform on WhatsApp that would allow users to connect additional AI agents. That points to a broader strategy than launching one assistant: the company may be trying to make its messaging ecosystem a distribution channel for multiple specialized agents, alongside its own products.

The second reported product, Watermelon, is expected in October. Its relationship to Meta’s model portfolio remains unresolved. The Decoder reported that it could join the Muse model family or be released as a standalone model. No technical specifications, capabilities or evaluation results were included in the source material.

Evidence and open questions

The strongest evidence in this report is media coverage of internal documents, not a public product announcement from Meta. That makes the launch timing, naming and feature list subject to change. “Coming weeks” is not a precise release date, and the October timing for Watermelon should be treated as reported planning rather than a confirmed ship date.

The commercial details are similarly unconfirmed. Meta is reportedly considering a tiered pricing structure for Hatch, including a premium subscription priced as high as $199.99 per month. The evidence does not say whether that price reflects a final plan, a proposed upper tier, or an internal scenario under review. It also does not identify which capabilities would be restricted to paid users.

There are no performance benchmarks, customer counts or adoption figures in the supplied sources. As a result, the report establishes Meta’s apparent product direction more clearly than it establishes whether Hatch can reliably complete tasks or attract paying users. Claims about the value of the agent will ultimately depend on execution, permissions, error handling and the quality of its integrations.

Watermelon presents a separate uncertainty. Without information about its size, training approach or intended use, it is not possible to assess whether the model represents a major technical release or a more targeted addition to Meta’s research and product portfolio. The possible connection to Muse is also unconfirmed.

Why a paid agent matters to Meta

Meta’s advertising business gives it a powerful distribution base, but the reported Hatch plan would test whether the company can sell direct access to AI capabilities. A subscription model could create revenue from users who want an assistant that performs useful tasks repeatedly, rather than from advertisers seeking audience reach.

The reported price range would also place Hatch in a different category from basic chatbot subscriptions. A high-end plan would need to offer dependable task completion, meaningful time savings or access to capabilities that users cannot easily reproduce with free assistants. If Hatch mainly routes users to existing websites without handling complex steps reliably, consumers may struggle to justify a substantial monthly fee.

For product teams, the more important issue may be the agent’s operating model. Connecting to Outlook, marketplaces, local-service platforms and social sites requires accurate identity management, permission controls and clear boundaries around actions such as purchases or messages. An agent that can browse but cannot safely complete tasks may have limited practical value; one that can act introduces risks around mistakes, fraud, privacy and user consent.

The WhatsApp platform could help Meta address distribution. WhatsApp already functions as a habitual communication tool in many markets, so adding access to third-party AI agents could reduce the friction of discovering and using them. It could also make WhatsApp a competitive layer between users and the services they rely on, although the supplied evidence does not explain how those agents would be reviewed, monetized or governed.

Implications for builders and enterprises

For AI builders, Hatch would reinforce the importance of integrations over standalone model access. The opportunity would shift toward agents that can maintain context, call external tools and complete workflows across fragmented services. At the same time, Meta’s reported plans could increase pressure on smaller agent companies to differentiate through specialized expertise, stronger reliability or better control over enterprise data.

Enterprise buyers should watch the separation between consumer convenience and business-grade automation. A system designed to plan trips or track fitness may not meet the auditability, access-control and data-retention requirements of a company. If Meta extends the same agent infrastructure into workplace use, buyers will need evidence about data isolation, administrator controls, logging and the handling of sensitive communications.

The reported pricing also raises a cost question for teams building with AI agents. A premium consumer subscription may be attractive if it replaces several narrow tools, but it could be expensive relative to task frequency. Developers will need to compare subscription fees with the cost of model calls, third-party APIs, human review and failed actions. Those economics cannot be evaluated yet because Meta has not disclosed a final plan.

What to watch next

The first signal will be whether Meta publicly confirms Hatch, provides a launch date and explains which integrations will work at release. Details about whether DoorDash, Etsy, Outlook and other services are official partnerships or agent-mediated access will clarify the product’s reliability and legal position.

Pricing tiers, usage limits and the capabilities reserved for a potential $199.99 plan will show whether Meta is targeting mainstream consumers or a smaller group of high-value users. Watch also for demonstrations that show complete workflows rather than isolated chatbot responses.

For Watermelon, the key signals are its relationship to Muse, the model’s intended users, independent evaluations and whether it is available through an API or only inside Meta products. On WhatsApp, developers and enterprises should look for documentation covering agent permissions, identity, privacy and platform review.

Creati.ai perspective

Meta’s reported Hatch strategy is less about another chatbot than about controlling a layer between users and online services. That makes integrations, trust and action reliability at least as important as the underlying model. The proposed subscription price suggests Meta believes a capable agent could support a premium market, but the available evidence does not yet show that Hatch can deliver the consistency required to earn it.

The next stage will be concrete product evidence: confirmed integrations, transparent pricing, safe execution and independent testing. Until Meta publishes those details, Hatch and Watermelon are best understood as reported components of a monetization strategy, not established products with proven market demand.