Why we built Roger

What problem does Roger solve?

Every marketer on your team now has an AI assistant. One uses Claude, another ChatGPT, the developer lives in Cursor or Claude Code. Each of them can read a Google Ads account, spot wasted spend and draft a better campaign. None of them can do it safely on their own.

To let an AI touch an ad account, someone has to wire up OAuth, MCP servers and developer tokens, then decide who may change what, how much it may spend, and how anyone finds out what happened. Every team we talked to rebuilt that plumbing, or gave up on it.

Roger is that plumbing, done once. You connect Google Ads, Meta Ads, Google Analytics, Search Console, Tag Manager and Merchant Center once per client. Then every teammate uses Roger from the AI they already pay for, through one MCP server.

Why a hub, and not another AI?

We started Roger as an AI marketer with its own chat. It worked, and it taught us something: people did not want another chat window. They wanted the assistant they already use every day to be good at marketing.

The models keep getting better on their own. What they lack is access to your accounts and the rules around it. So Roger stopped competing with Claude and ChatGPT and became the layer they plug into: the connections, the tools, the playbooks (we call them skills) and the safety net.

Is it safe to let an AI change an ad account?

Only with rules the AI cannot talk its way around. In Roger they are checked in code, outside the model:

• Guardrails per client: budget caps, protected campaigns and other rules you pick. A change that breaks one is blocked and logged.
• Drafts first: a change becomes a draft you approve before it reaches the account, and new campaigns start paused.
• Roles per client: Admin, Editor or Viewer, so a teammate can analyze one client without being able to edit another.
• An activity log: who asked, from which AI, what changed and why. Never the prompt itself.

That is the part we would want before handing any system the keys to an account we spent years building.

Built by advertisers

Between Jean and me, we have spent hundreds of thousands of dollars on Google Ads, managed dozens of accounts and built several SaaS products along the way. We know the tedium of weekly audits and the risk of a budget running unchecked overnight.

When Claude Code came out, we saw what AI could do for PPC. We also saw every agency hit the same wall before a single ad got better. Roger is what we wished we had on day one.

What it costs

A flat $25 per client per month, with unlimited teammates, connectors and AI assistants. The trial lasts 14 days and needs no card.

What’s next

We keep adding skills and connectors. If you want to try it, run the Google Ads audit from your own AI, or see how it works.