Meta launches Muse – WhatsApp AI agent with integrated Stripe payment, email and price‑negotiation features

Meta announced the AI agent Muse on September 10, 2026 – a virtual assistant that communicates exclusively through WhatsApp and handles tasks that previously required human interaction. Muse can autonomously fill out forms, book travel, negotiate prices and – notably – make payments via Stripe’s Link service. It also drafts emails and sends them after user approval. The launch marks the first public rollout of an AI that not only advises but actively executes transactions on the user’s behalf.

How Muse works

The agent runs on an isolated virtual machine managed by Meta. Users launch Muse via a WhatsApp chat and issue commands such as “Book me a flight to Berlin for October 15” or “Buy the new smartphone I saw on sale.” Muse then opens a browser, navigates to the relevant sites, fills in the required forms and, if necessary, conducts price negotiations. The entire process is overseen by a security agent called Sentinel, which checks every outbound request before it reaches the internet.

The interaction remains transparent: Muse notifies the user whenever a step requires human approval – for example before the final purchase or before sending an email. This keeps control in the hands of the user while the actual execution runs automatically.

Payment processing via Stripe Link

For payments, Muse uses Stripe Link, a service developed by Stripe that enables one‑time card payments. After user confirmation, Muse generates a single‑use, tokenized payment link that does not expose sensitive card data. The user sees only the amount and recipient information before authorising the transaction. This approach meets common security standards and reduces the risk of data misuse.

Using Stripe Link sets Muse apart from other AI providers, which have either lacked an integrated checkout function or have subsequently disabled their payment services. By combining a chat interface with instant payment capability, Meta positions itself as a pioneer in agentic e‑commerce solutions.

Competition, data protection and regulatory implications

The market for generative AI is fiercely contested. OpenAI recently disabled the direct checkout feature in ChatGPT for regulatory reasons, while Google and Microsoft have yet to offer a comparable end‑to‑end solution on their messaging platforms. Muse could therefore gain a competitive edge, especially in the German market where WhatsApp remains the dominant messaging tool.

At the same time, the autonomous execution of transactions raises data‑protection and consumer‑safety concerns. The European Union adopted the AI Regulation (AI Act) in December 2025, which tightly regulates especially “high‑risk” AI systems. Because Muse operates independently on the internet and carries out financial transactions, it is likely to fall under the stricter obligations – such as the requirement for transparent documentation of decisions and the provision of grievance mechanisms.

Meta stresses that Sentinel reviews every action and asks the user for approval when uncertainties arise. Nevertheless, it remains to be seen how regulators will assess the practice. One potential scrutiny point is whether the AI’s decisions are traceable and explainable when it conducts price negotiations or alters contract terms.

Quick overview: Muse compared with existing AI agents

The table shows that Muse is currently the only solution that combines all listed functions in a single WhatsApp‑based agent. Whether this combination remains permissible under regulations in the long term will be revealed in the next phase of the rollout.

For businesses and consumers, Muse represents a step toward fully automated digital assistants that not only inform but also act. The integration of payment services and the ability to conduct price negotiations could change how online purchases and service requests are handled – provided that security and data‑protection standards meet user expectations.