TrendsFast for marketing agencies: one brand context per client

Who this is for

Agencies and freelancers who deliver content, social or growth work for several clients and are building agent workflows to do it: research in Claude or ChatGPT, a script that prepares the weekly brief, an internal tool. The bottleneck is rarely writing. It is knowing, per client, what their customers are discussing this week, with sources you can put in front of the client.

Keep every client’s context separate

• One workspace per client brand. Each has its own confirmed context: the audience, the communities they gather in, the competitors, the markets. Research for one client is ranked against that client only.
• Keys that cannot cross over. An API key belongs to one workspace, and the server decides the workspace from the key; naming another client’s website in a request does not reach its data.
• Saved results per workspace. Each client’s results can be reread for free, so last month’s evidence is still there for the report.
• Brands per plan. Free, pay-as-you-go and Starter hold 1 brand; Growth holds 3 and Scale 10. Credits are shared across the account.

A repeatable monthly routine, priced

An example of one client’s month, using each product when its question comes up. It is an illustration of the arithmetic, not a measured usage pattern.

That is 225 credits per client a month. Three clients use 675 of Growth’s 10,000 monthly credits ($99 a month); ten use 2,250 of Scale’s 40,000 ($299). A source read in the last few hours is reused at no cost and an empty result costs nothing, so a real month can cost less.

One workflow, run for every client

The same script serves every client because the key carries the client. This loop asks each client’s Customers question and keeps every result’s ID for the report. In Claude or ChatGPT, the equivalent is one MCP connection per client workspace.

// Node.js 22+, server-side. One key per client workspace, each in its own variable.
const clients = [
  { name: "client-a", key: process.env.TRENDSFAST_KEY_CLIENT_A },
  { name: "client-b", key: process.env.TRENDSFAST_KEY_CLIENT_B },
];
for (const client of clients) {
  const headers = { authorization: `Bearer ${client.key}`, "content-type": "application/json" };
  // The key decides the workspace: this context is that client's, and no other.
  const { context } = await (await fetch("https://trendsfast.com/v3/trends/context", { headers })).json();
  const requestId = crypto.randomUUID(); // store per client before sending
  const response = await fetch("https://trendsfast.com/v3/trends/niche", {
    method: "POST",
    headers,
    body: JSON.stringify({
      request_id: requestId,
      max_credits: 10,
      input: { product: "customers", context_id: context.context_id },
    }),
  });
  const result = await response.json();
  console.log(client.name, result.result_id, result.data?.brief?.headline);
}

What TrendsFast does not do for agencies

• No team seats, client roles or client logins: every brand belongs to the account that created it.
• No white-label reports and no per-client invoices. Credits come from one balance; track each client’s spend from its own workspace’s history.
• No posting, scheduling or approvals. Drafts your agent makes from the evidence are yours and your client’s to review.
• Coverage follows each client’s confirmed communities, competitor accounts and categories; a niche the index barely covers returns less, and says so.

When an agency should use something else

If you need a social listening suite with seats, alerts and client dashboards, or a scheduler with approval flows, use one; TrendsFast can still feed it research. If your clients’ work is mostly short-form video, a video intelligence platform goes deeper on formats, hooks and creators.