Agency Mode

Run several client brands from one account: centralize AI billing on one shared key while keeping each client's publishing isolated.

What you'll learn

  • How to structure m18t when you run several external clients.
  • How to centralize AI billing while keeping each client's publishing isolated.
  • How roles and brand scope control what each member of your team can reach.

m18t started as a tool for founders running several of their own projects. The same multi-brand shape works well for agencies running operations on behalf of external clients.

"Agency Mode" isn't a toggle in settings. It's a way of arranging your workspace: one account (business), one brand per client. You operate everything from inside; the clients see the finished output on their own websites and social accounts, not inside your studio. (For the account-vs-brand layering, read Brand vs. Business.)

Centralize AI billing, isolate everything client-facing

Share one AI key across clients

Agencies usually absorb AI costs rather than chasing fifteen clients for fifteen OpenAI keys.

  1. Generate one API key from your agency's own provider account.
  2. Add it once on the Vault tab of the Connections page.
  3. Point each client brand's AI connection at that same vault key.

You bill in one place; the generated content still stays separated by brand. See The Secret Vault.

Keep social and CMS targets per-brand

Publishing destinations are the opposite — they must never be shared. Sending Client A's post to Client B's Facebook page is the kind of mistake that loses an account.

m18t's brand-scoping guards against this: while Client A's brand is active, the studio only sees Client A's channels and credentials. Configure each client's Meta, CMS, and other targets while that client's brand is the active one in the Brand Switcher. See Per-Brand Integrations.

Scope automations deliberately

If you run a cron job, you can scope it to a single brand. Set the active brand before creating brand-specific scheduled work so it lands where you intend.

Team access and brand scope (read this before onboarding anyone)

Multi-user is live: you can invite members, and two controls decide what each one reaches. A role governs what a member can do. A brand scope governs which brands they can see and act on — set it per member on the Team page.

A member with no brand scope set can access every brand in the account. Unrestricted is the default until you set the scope — so set it when you invite, not after.

The practical consequence for agencies: scope every login that shouldn't span clients. Your own operators can stay unrestricted if they genuinely work across brands; anyone else — including a client you invite to review their own brand — gets a brand scope of exactly their brand and an appropriately limited role. Many agencies still keep logins to their own operators and share finished output externally: a scoped member does see work-in-progress inside their brand.

Pricing implications

Pricing isn't finalized in beta, but the working assumption is that m18t meters on brands per account, not on seats. That's deliberately friendly to agencies: adding an operator to your team isn't what scales your bill — adding client brands is. And you pay your AI provider directly for the raw token cost; m18t doesn't mark it up. Treat the brands-per-account axis as the direction, not a committed price — see Brand vs. Business.

FAQ

Can my clients log in to see their dashboards? Yes, with care. Invite the client as a member, set their brand scope to exactly their brand, and give them a limited role — at invite time, because an unscoped member can access every brand. Whether to grant client logins at all is a workflow call: a scoped member still sees work-in-progress in their brand, so many agencies keep logins internal and share finished output externally.

How do I stop publishing one client's content to another's account? Brand-scoping handles it: only the active brand's channels and credentials are visible at any time. The discipline is to confirm the Brand Switcher shows the right client before you configure or publish.

Can all my clients share one AI key? Yes — that's the recommended setup. One agency key in the vault, each client brand's AI connection pointing at it. Each client can still have its own if a client insists on their own billing.

Is there an extra fee per client brand? Creating a brand triggers no infrastructure or billing event in itself. Longer term, brands per account is the expected metering axis, so client brands are the dimension that scales your plan. Beta pricing isn't locked.

Can I assign work to specific operators? Yes. Content, artifacts, emails, events, assets, and PD entries can be assigned to a member, and the assignee is notified. Combine assignment with brand scope so each operator's queue matches the clients they cover.

What's next

Was this article helpful?