Skip to main content

Automation overview

MessageGears delivers a family of automation-focused REST APIs for composing and launching platform workflows without going through the Accelerator UI — by script or pipeline calling the API directly, AI agents driving the same workflow on your behalf through Model Context Protocol (MCP). A client authenticates with its own API key, works through the same operations available in the UI, and the assets it creates or changes appear in Accelerator like any other.

In Accelerator 26.3.2, external campaign APIs deliver agent-ready endpoints for automation. More endpoint groups and use case articles will be added with future releases.

Why MessageGears is building this layer​

More of what happens on the platform is driven by client-developed tools, orchestration layers and AI agents acting on a marketer's behalf, as an accompaniment to a marketer performing tasks through a UI. MessageGears is building a REST API layer first — a well-defined, permissioned interface that any automated caller can use, whether that caller is a script your team wrote, a pipeline step, or an AI agent.

That same layer is the foundation for MessageGears' upcoming MCP server. MCP introduces a new way to reach the capabilities this automation layer already exposes, by letting an AI agent call these APIs as tools in a conversation instead of you calling them directly. Building the API surface first means every capability automation adds is immediately available to a script today and to an MCP-connected agent later.

How automation works​

Every automated workflow, regardless of which API it calls, breaks down into the same three participants:

  • You - State the outcome you want, in plain English to an agent or as structured input to a script: an audience, a destination, a schedule.
  • The automation - A script, pipeline, or AI agent that turns your request into the right sequence of API calls.
  • MessageGears - The marketing execution and delivery platform. Validates every call, applies the change, and enforces guardrails before anything consequential — like a live send — takes effect.

Underneath, automation APIs work the same way regardless of which asset type they manage: they let a caller create, read, update, and delete the asset through standard REST operations, scoped to whatever permissions the caller's API key holds. A script or agent can create a new asset, read back its current state to verify or branch on, update one part of it without resending the whole thing, or remove it — the same four operations you'd otherwise perform by hand in the UI.

What's available today​

  • External campaign automation - Compose and launch External Campaigns, syncing an audience to destinations like Meta Ads, Google Ads, SFTP, or Amazon S3, without configuring through the platform UI. For more information, see External campaign automation overview.

What's coming​

Automation coverage is a key focus of MessageGears and will expand over time. Model Context Protocol support is next: an MCP server that exposes these same APIs as tools an AI agent can call directly in conversation, without you writing or running client code yourself.

Practical use case​

Launching a paid media campaign through an AI agent

A marketer wants to launch a paid social campaign that retargets high-propensity customers — people MessageGears' predictive models score as likely to convert — across Meta and Google Ads. Instead of building the campaign by hand, they describe the goal to an AI agent connected to MessageGears: create a campaign, set the audience and adjust the segmentation variables, and launch on a set schedule to delivery the qualified audience to both ad platforms.

The agent turns that request into a sequence of API calls: it looks up the audience source that already surfaces the propensity score, creates a campaign draft, attaches the audience and both destinations, adjusts variables, sets a schedule and requests activation. MessageGears validates and the resulting campaign appears in Accelerator exactly as if a person had built it in the UI, ready to monitor like any other campaign.

Prerequisites​

Every automation API authenticates with an API key, and every key is scoped to the actions it's allowed to take and, where the API supports Brands, the Brands it can access. Before building against any automation API:

  • Create an API key with the right scope - A key inherits its creating user's role and, for APIs that support Brands, the Brands selected at creation. Only request or grant a role broad enough for what the integration actually needs to do. For more information, see API key management.
  • Confirm Brand access, if you use Brands - If your instance uses Brands to separate campaigns or assets, make sure the key's Brand association covers everything the automation needs to touch — Brand access can't be changed on an existing key.
  • Check the specific prerequisites for the API you're calling - Each automation API has its own entities and ids you'll need before your first call.

Next steps​