The daily publication · November 10, 2026 · Small business
Meta builds a sales team to sell its AI agents to small businesses, while big retailers refuse to hand it their customers

Meta is building a sales team to sell its AI agents to small businesses.
The Information reported on November 9 that Meta has formed a sales organization to sell its AI products. One team focuses on small and mid-sized businesses. Consulting partnerships with Accenture, Deloitte and IBM are meant to reach larger corporate clients. PYMNTS and Daily AI Brief both reported the move the same day, citing The Information, a source familiar with the matter, and Meta job postings.
The push sits on the Meta Enterprise Platform, launched in September. That platform carries Muse Code, business agent software and Muse Spark models. It is headed by former MongoDB CEO Chirantan Desai, who goes by CJ. Mark Zuckerberg called the platform the company next major pillar.
Meta plans capital spending of up to $145 billion this year, much of it for AI. This sales team is one way that bill starts getting paid.
The test market is already there. PYMNTS notes Meta saw strong demand from small businesses in 2024 for its automated customer agents, the ones that manage inquiries around the clock on WhatsApp, Messenger and Instagram.
The idea
Your customers already talk to your business inside Meta apps. Now Meta wants to sell you the agent that does the talking back.
That changes the question. It is no longer, should I use an AI agent to answer customers. It is, whose agent is this, what data does it see, and who owns the relationship when the agent answers in my name.
An agent is not a chatbot on a page. It is a worker you let speak to customers when you are not in the room. The workflow around it matters more than the demo. What it can promise. What it can discount. When it must stop and hand the conversation to a person. Those rules are yours to set, or they get set for you.
The example
Look at the big retailers. They got the pitch first, and they pushed back.
EMARKETER reports that Amazon and Booking.com hesitated over Meta agent pitch in October. Their concern was control: how agents interact with customers, and whether they trust Meta with access to user and operations data. Meta ultimately discontinued its push to large retailers.
That is not a small business story yet. It is a warning written in plain sight. The largest companies with the most leverage looked at the same offer and asked who controls the agent, who sees the data, and who owns the customer. A ten person shop has less leverage, not more. So it has to ask those questions earlier, in writing, before it signs.
EMARKETER puts the conclusion plainly. Brand control over agent behavior, voice, data access and human escalation may decide whether businesses adopt these agents at all, more than the technology itself.
There is also history here. PYMNTS notes Meta Workplace software peaked at 7 million paid subscribers in 2021 and was discontinued in 2024 after struggling against Microsoft and Slack. A platform can be big, well funded and still get shut down. An owner who builds a customer workflow on it needs an exit plan from day one.
The next step
Before any agent answers a customer in your name, write down four controls on one page.
First, what the agent can say and do on its own, and what it can never promise, discount or refund without you.
Second, what data it can see. Customer names and order history are one thing. Your full operations data is another. Start with the smallest access that still does the job.
Third, when it must hand off to a person, and how fast that person gets the full conversation, not a summary.
Fourth, how you get your data back and turn the agent off if the product changes, the price jumps or the platform gets discontinued.
Then test it on one channel, with real questions from your own customers, for two weeks. Read every conversation. Count the handoffs. Count the promises it made that you would not have made. That count is the real price, no matter what the subscription says.
And hear the other side clearly. Big retailers refused to hand Meta their customers, over control and trust in the data. That refusal is not proof the agents do not work. It is proof that control is the product. If a vendor cannot answer the four controls above in plain words, you do not have an agent problem. You have a trust problem, and no discount fixes that.
Want help finding the one bottleneck in your operation worth fixing first? A free 10-minute workflow conversation will map where the hours go and who signs off before anything final. No pitch, just the map.
Sources: PYMNTS, https://www.pymnts.com/news/artificial-intelligence/2026/meta-forms-sales-group-to-sell-ai-to-small-businesses/; EMARKETER, https://www.emarketer.com/content/meta-pushes-enterprise-ai-adoption-businesses-guard-customer-data-loyalty; Daily AI Brief, https://dailyaibrief.com/news/meta-builds-ai-sales-team-for-businesses-vgzIIXhJ. Opposing view: EMARKETER, https://www.emarketer.com/content/meta-pushes-enterprise-ai-adoption-businesses-guard-customer-data-loyalty; PYMNTS, https://www.pymnts.com/news/artificial-intelligence/2026/meta-forms-sales-group-to-sell-ai-to-small-businesses/.
Sources
The other side
The resistance is already on record. EMARKETER reports Amazon and Booking.com hesitated over Meta agent pitch in October, in part over control of how agents interact with customers and whether they trust Meta with access to user and operations data, and Meta ultimately discontinued its push to large retailers. PYMNTS adds the precedent that Meta Workplace software peaked at 7 million paid subscribers in 2021 and was discontinued in 2024 after struggling against Microsoft and Slack. EMARKETER concludes brand control over agent behavior, voice, data access and human escalation may decide adoption more than the technology itself.
Source: EMARKETER
Connected stories
"A three-person company gave an AI agent the marketing job. Late invoices fell from 56% to 8%."
Same company: Meta; Same topic: trust and verification; Same topic: AI assistants
"Goldman will teach owners AI. Only 15% have actually wired it in"
Same company: Meta; Same topic: trust and verification
The AI that finishes the job, not just drafts it
Same topic: trust and verification; Same topic: AI assistants
"The 6 a.m. sick call just got an assistant"
Same topic: trust and verification; Same topic: AI assistants
Want this every week?
One idea, one example, one next step. Free, every Wednesday. Or step into the rooms where owners talk it through.