Opteo gives you two ways to work with AI on your ad accounts.
Agent lives inside the Opteo app. See Agent.
The Opteo MCP connects Opteo to your own AI assistant, such as Claude or ChatGPT. To set it up, see Opteo MCP.
What both can do
Both can read your performance data and make changes on Google Ads, Microsoft Ads, Meta Ads, LinkedIn Ads and TikTok Ads, with your approval. Both can also read GA4, search the web, research competitors and work with files you upload. For the full list of changes on each platform, see Make changes with Agent.
Video campaigns on Google Ads can't be changed by either, because Google doesn't allow it through its API.
Where they differ
Which ad accounts they change: Agent changes only the ad account the conversation belongs to. The MCP can change any ad account you've linked in Opteo, so it suits work across many clients at once.
How you approve changes: Agent shows a confirmation card in the app. On Google Ads you can edit the details on the card first. Through the MCP, your AI assistant asks you to confirm before it runs anything. How that looks depends on your assistant. In Claude, for example, you can allow each change or always allow it. We recommend keeping confirmations switched on.
Opteo settings: the MCP can manage your Opteo setup, such as linking and unlinking ad accounts, changing ad account settings, organising account groups and saving prompts. Agent doesn't do this yet.
Improvements and Issues: the MCP can list Improvements and Issues, push and dismiss Improvements and recheck Issues. In Opteo, you do this from the Improvements and Issues pages.
Images: Agent can generate images, for example to help with new ad creative.
Where your conversations live: Agent conversations are saved in Opteo. MCP conversations stay in your AI assistant. Both use the same saved prompts and team prompts.
Changes go live as soon as they're approved, so check the details first.
Need something that isn't listed?
If there's something you'd like Agent or the MCP to do, get in touch with our support team. We'd love to hear what you need.