How it works

The MCP architecture: a secure bridge between your assistant and the ad platforms.

Model Context Protocol (MCP)

MCP is an open standard for connecting AI assistants to external systems. Opus Growth is a remotely hosted MCP server: once you add the connection URL to your assistant, it can call the tools we provide.

# Flow
You  →  Claude/ChatGPT  →  Opus Growth (MCP)  →  Ads: Google / Meta / TikTok / LinkedIn / Microsoft
       # natural language        # secure tools       Data: Search Console / Analytics / YouTube / Business Profile / Tag Manager
                                            # all official APIs

Identity and session

When your assistant connects to the MCP server, you sign in with Google. You authorize each ad platform (Google, Meta, TikTok, LinkedIn, Microsoft) once; your access tokens are stored encrypted and are tied only to your account.

Google's data sources — Search Console, Analytics, YouTube, Business Profile, Tag Manager — are authorized through separate connection flows and each is stored with its own encrypted token. Thanks to this separation, connecting or disconnecting one source doesn't touch your Google Ads connection. The cross-channel tools bring all connected sources together into a single marketing funnel.

How a request becomes an API call

You write a sentence; the assistant picks a tool and fills its arguments; we call the platform's official API with your token and hand the result back. Nothing is scraped and no browser is driven in the background, which is why an operation either succeeds or returns the platform's own error message rather than failing in some ambiguous way.

Two consequences are worth knowing. First, platform rules are the platform's, not ours: if Google rejects a headline for being too long, that limit is Google's and we surface it verbatim. Second, whatever the API cannot do, we cannot do either, and where that bites we say so on the relevant page instead of leaving you to discover it.

Why you do not see every tool at once

There are hundreds of tools across the platforms, and handing an assistant all of them at once makes it slower and less accurate at picking the right one. So the set is scoped: the relevant tools for what you are doing are exposed, and the rest stay out of the way until they are needed. You never lose access to a capability by not seeing it listed.

Built-in safety

Every write operation (creating a campaign, changing a budget, pausing an ad) is two-step: first a preview, then confirmation. Budget spikes are also blocked, and every action is written to the audit log. Details: Security & Confirmation.

Common questions

Does my ad data get used to train an AI model?

Not by us. We call the ad platforms' APIs with your token and return the result to your assistant; the data is not collected for training and is not shared with other accounts. What the assistant you chose does with a conversation is governed by that vendor's own policy, which is worth reading if the question matters to you.

Is this an official Google, Meta or Microsoft product?

No. Opus Growth is an independent connector built on those platforms' public APIs. Your account stays yours, the authorization is yours to revoke, and nothing here replaces the platforms' own interfaces.

What happens if a platform is down or rate limits me?

You get the platform's error passed through rather than a made-up answer, and where a call was already billed it is refunded. Retrying later is usually the right move; a rate limit is a statement about the platform's quota, not about your account.

Do I need one subscription per ad platform?

No. One Opus Growth account covers every platform, and each is connected once. The plan governs how many actions and how many accounts, not which platforms you are allowed to touch.