News at a Glance
The tech giant says the agent can help owners run their business and find new customers.
What Happened
On September 29 Meta announced through its corporate blog and the Meta Business Suite help documentation that Muse, its AI agent for commercial use, would move beyond a limited advertiser pilot to small and mid-sized businesses. Muse had previously been available to a subset of North American advertisers for ad creative generation and campaign optimisation; after the expansion, merchants can invoke it directly inside WhatsApp Business, Messenger, Instagram Direct and the Facebook Page inbox to auto-answer customer questions, organise order and booking details, and tag customers by purchase intent. Meta listed the first wave of capabilities as multi-turn conversational rep
Background
Automation for small merchants is not a new demand, but for the past two years it has been the territory of SaaS vendors: Salesforce's Agentforce bills per conversation, HubSpot has written AI into its CRM, and Shopify has embedded chatbots in store back ends. Meta's situation is different. Its advertising business is close to a ceiling, while WhatsApp Business messaging is among its fastest-growing segments. Three preconditions made this the moment to move. First, the open-source Llama family has pushed inference costs low enough that a free conversatio
Deep Dive
Liu Gong argues the interesting question about the Muse expansion is not what the agent can do but what Meta intends to sell. From a technical standpoint, Muse's scope has been deliberately narrowed: multi-turn replies, asset generation, lead qualification and human handover, all grounded in the merchant's own catalogue and conversation history rather than open-ended general reasoning. That is a pragmatic call. Small merchants have high conversation volume and heavily repeated intents, so retrieval templates plus a fine-tuned small model cost far less at the margin than a frontier model, which lets Meta give away a base quota per conversation and still keep the cost inside advertising gross margin. The trade-off is that Muse cannot handle cross-language after-sales issues or bespoke product questions, and those cases still fall back to a human. On the business side, what Muse changes is how Meta charges. Meta has sold impression-priced ad inventory, so revenue rose and fell with merchant budgets; Muse lets it price on operational outcomes for the first time, whether that is qualified leads, booked appointments or response time. Salesforce already lists Agentforce at roughly two dollars per conversation, which gives Meta a ready anchor: as long as per-conversation acquisition cost undercuts hiring a support person, Muse has room to carry a premium. The more important effect is data flowing back. Catalogues and customer preferences that merchants upload will feed ad targeting efficiency, letting the advertising and subscription curves lift each other. For the industry, the sharpest pressure lands on small-business SaaS and outsourced customer support. Once a merchant already runs the business inside WhatsApp, charging them again for a second back office becomes a much harder sell. Liu Gong expects two kinds of response over the next twelve months: SaaS vendors wiring their own agents into Meta's business APIs to add value rather than be replaced, and vertical support outsourcers being forced to cut prices. One caveat matters more than model quality: whether Meta can hold the line on not using merchant customer data to improve its own ad models. The moment that boundary is questioned, merchants will move to their own sites or third-party CRMs faster than most people expect.
Data Points
- Meta's business tools reach roughly 200 million small and mid-sized merchants, against a test-phase base of a few hundred thousand weekly active business-AI users, so the expansion takes availability from under 1% of that pool to essentially all of it.
- Muse's base quota ships free with a Business Suite account with overage billed per conversation, while Salesforce's Agentforce lists at about $2 per conversation, pushing Meta's viable band toward $0.10 to $0.
- – Meta's family of apps reports about 3.4 billion daily users, with WhatsApp above 3 billi
Impact & Outlook
For the industry, the most immediate change within three months is a lower pricing reference point: per-seat subscriptions aimed at small merchants from Salesforce and HubSpot will face comparison pressure, and quotes from outsourced support and entry-level agency operators will be hit first. Within a year, merchant operations will concentrate further inside Meta's ecosystem, forcing standalone storefront builders and third-party CRMs to prove their value on data ownership and cross-platform reach or risk becoming add-ons to Muse. For ordinary users, the
Angles to Explore
Points to consider
- How Meta's advertising and subscription curves reinforce each other once the agent shifts from an ad tool to an operating subscription
- How per-conversation pricing reshapes the small-business SaaS and outsourced support markets
- Whether merchant customer data feeds Meta's ad models, a boundary that will determine how quickly merchants migrate away
Source & Original
This item comes from TechCrunch AI (published 2026-09-29 13:47:30). Wenfeng Radar provides a Chinese/English summary and commentary on overseas AI developments; the original content is owned by its author.
Further reading: 2026 Frontier LLM Comparison: A Practical Guide — a durable, evergreen reference. Bookmark it.
Daily aggregation of overseas AI developments with in-depth analysis. Bookmark us to never miss an important signal; back to homepage.