Meta Muse for Small Business: What the New AI Agent Can Actually Do

Meta Muse for Small Business is an AI agent, announced September 29, 2026, designed to work across the business tools a small business already uses rather than only answer questions about them. It can read data from connected apps such as Shopify, Stripe, QuickBooks, HighLevel, Klaviyo, and Slack, plus a business's own Instagram, Facebook Page, and Meta ad accounts, and it can draft content, plan work, and take action inside those tools — but Meta says nothing publishes, sends, or spends without the owner's approval. The base Muse app has already rolled out in the US and Canada, which makes today's small-business expansion immediately relevant to businesses in both countries, though Meta's own announcement doesn't give a separate, explicit country list for the small-business feature set specifically — verify inside your own Muse app settings rather than assuming.

What Changed on September 29, 2026

Meta's base Muse app launched September 8, 2026, as a personal AI agent capable of completing multi-step tasks — including, as covered in our companion Holiday Marketing 2026 article, a consumer shopping-and-checkout path through Shopify. Muse for Small Business is a distinct expansion announced three weeks later: instead of a consumer delegating a shopping task, a business owner connects Muse to the tools that run their actual operations — sales, payments, bookkeeping, scheduling, marketing, and creative work — so the agent has context across a business's whole operation rather than acting as a standalone chatbot. Meta also brought in Chirantan "CJ" Desai, MongoDB's former CEO, to lead this broader enterprise push, a leadership signal that this is meant as an ongoing platform investment, not a one-off feature.

Where It's Available

The underlying Muse app has already reached the US and Canada — reporting around its September launch describes it topping app charts in both countries. Meta's own September 29 announcement of the small-business feature set doesn't, in the sources reviewed for this article, publish a separate country list specific to those new business skills and connectors. The reasonable reading is that Muse for Small Business rides on the same US/Canada footprint as the base app, but that isn't the same as an explicit confirmation — check your own Muse app settings directly before assuming access, and don't plan a launch or campaign around a market Meta hasn't confirmed for this specific feature set.

The Confirmed Connector List

Meta says the full, current connector list lives inside the Muse app's own settings, that custom connectors can cover services not yet natively supported, and that other companies can apply to be added — meaning this table is a snapshot of launch day, not a permanent ceiling. A business should check its own app for the current list rather than treating this article as the definitive, unchanging set.

What It Can Do Today

• Read and analyze: Instagram and Facebook Page analytics, ad account performance, and data inside connected apps (sales figures in Shopify, invoices in QuickBooks, tasks in Asana, and similar) — the specific depth of read access depends on the connector and what that app's own API exposes to Muse.
• Draft: content ideas, campaign concepts, task lists, meeting summaries, and messages — drafting is explicitly lower-risk than executing, and Meta's own framing treats it as the safe default action.
• Plan and coordinate: pull context from multiple connected tools to propose next steps, similar in spirit to how a human assistant would cross-reference a calendar, a task list, and a CRM before suggesting what to do next.
• Execute, but only with approval: publishing content, sending messages, and spending money are the three actions Meta explicitly says require the owner's go-ahead before Muse can carry them out.

What Still Needs Approval — and How Granular That Approval Can Be

Meta describes five approval types for Muse: one-time (approve this specific action once), session (approve for the current working session), task (approve for the duration of a specific multi-step task), time-bounded (approve for a set period), and perpetual (approve indefinitely, until revoked). This matters more than a simple "ask before every action" framing would suggest — it means an owner can, in principle, grant Muse standing permission to draft weekly social captions without asking each time, while still requiring a fresh approval every time it wants to send a customer message or spend ad budget. Whether that granularity holds up in practice, connector by connector, is something only using the product will confirm; this article is describing Meta's stated design, not an independent test of it.

CRM and Workflow Implications

HighLevel — a genuine CRM and marketing-automation platform — is on the confirmed connector list, which is the most directly CRM-relevant fact in today's announcement. But no source reviewed for this article specifies exactly what Muse can do inside HighLevel: whether it can only read contact and pipeline data, or whether it can also create and update records, trigger a workflow, or move a lead through a pipeline stage. Treat "connects to HighLevel" and "can write to your CRM" as two different claims until Meta's own documentation is more specific — a business relying on this connector for actual lead handling should verify the exact permission scope inside its own Muse settings before building a process around it, not assume full read/write access from the connector's mere existence.

The practical workflow, based on what's confirmed: a business goal ("follow up with leads who inquired this week") → Muse pulls context from connected tools (HighLevel contact records, recent Instagram DMs, ad account performance) → Muse proposes a draft action (a follow-up message, a task assignment) → the owner approves at whatever granularity they've set → Muse executes → the result should still land in the actual system of record (the CRM, the accounting software) rather than living only inside a Muse conversation.

Advertising and Campaign Use Cases

Because Muse connects directly to a business's Meta ad accounts alongside Instagram and Facebook Page data, it's positioned to analyze campaign performance and social engagement together and draft next-step recommendations — a promotion idea informed by what's already resonating organically, for example. This is meaningfully different from Meta Business Agent (a separate, earlier Meta AI feature covered in our Meta One article that answers customer questions and qualifies leads inside a conversation) — Muse for Small Business is about the owner's own operational and planning work, not customer-facing conversation.

Ecommerce and Finance Use Cases

With Shopify and Stripe both connected, Muse can plausibly analyze sales and payment data together and summarize campaign or inventory performance in plain language — useful for a business owner who doesn't want to open three dashboards to answer "how did this week go." QuickBooks adds a bookkeeping-adjacent read: summarizing expenses or flagging anomalies is a reasonable extrapolation from what's confirmed, though this article isn't claiming Muse can file taxes or make accounting judgments a bookkeeper should still make.

Privacy, Permissions, and Governance

• Credential isolation: Meta says Muse has no visibility into passwords or payment methods — credentials go into secure storage that Muse can use to act without ever seeing the underlying credential itself.
• Audit trail: Meta says Muse shows a complete record of what it has done and what it plans to do next, giving an owner a way to review its activity rather than trusting it blindly.
• Ads-data separation: Meta says a person's Muse conversations and connected-tool data aren't shared with Meta's own advertising systems — a distinct claim from whether Muse itself can see and act on a business's ad account data, which it explicitly can.
• Per-connector control: a business chooses which apps Muse connects to and how much access each one gets, and can change or revoke that access at any time.
• Opt-out of AI training: Meta says users can opt out of having their interactions used to train its AI models.

Practical Examples by Industry

• Ecommerce: Muse analyzes recent Shopify sales and ad performance together, drafts a promotion concept, and proposes next steps — the owner still approves before anything publishes or spends.
• Real estate: Muse organizes inbound lead information (where a genuine CRM connector exists), summarizes recent inquiries, and drafts outreach messages for the agent to review and send.
• Professional services: Muse drafts a proposal outline from a Notion brief, summarizes a Zoom meeting, and coordinates related Asana tasks — reducing the mechanical part of pulling those pieces together manually.
• Home services: Muse drafts follow-up messages for recent jobs and organizes upcoming work from connected scheduling/task tools, with a human still sending the final message.
• Marketing-focused teams: Muse reviews Instagram/Facebook and ad-account performance together and drafts the next round of content or campaign ideas from what's actually working.
• Operations: where supported, Muse can review expenses, calendars, and tasks across connected tools to surface what needs attention — a triage assistant, not an autonomous decision-maker.

None of this amounts to fully autonomous operation. Every example above still runs through the same shape: Muse proposes or drafts, a human approves the actions that matter, and the underlying systems of record — the CRM, the accounting software, the ad account — remain the actual source of truth, not the Muse conversation itself.

Where Meta's Copilot-Style Competitors Fit — and Don't

Muse for Small Business's closest conceptual competitor isn't a chatbot — it's Microsoft's Copilot agents (covered in our separate Copilot article), which similarly connect to a business's existing Microsoft 365 tools rather than operating standalone. The meaningful difference is ecosystem: Copilot's natural reach is Microsoft's own productivity suite, while Muse's confirmed connectors lean toward commerce, payments, and the specific small-business tool stack (Shopify, Stripe, QuickBooks, HighLevel) rather than office productivity. A business already committed to one ecosystem will likely get more out of that platform's own agent than trying to bridge both.

Where NextFlow Fits

Connecting Muse to your tools is the easy part — the harder, more valuable work is making sure what it drafts and proposes actually flows into a CRM built for follow-up, not just a chat history. NextFlow's AI Solutions, CRM & Automation, Lead Generation, and Website & Landing Pages services are built for exactly that gap: turning an AI-assisted draft or a qualified lead into something your business actually tracks and acts on.