Ad Agencies Are Building Their Own AI Tools in a Single Evening. Here's Why That Matters.

A VP Built a Monitoring Platform Before Breakfast

Mitch Hislop, VP at ad agency Broadhead, sat down one evening with Claude Code and built his agency's entire GEO monitoring platform. Not a mockup. Not a pitch deck. A working tool. By the next morning, his team had something that would have taken months to spec, bid out, and build the traditional way.

He's not alone. Agencies like Havas, Broadhead, and Supergood are skipping the "build vs. buy" debate entirely. They're just building.

What They're Actually Building

The hot category right now is GEO, or Generative Engine Optimization. Think of it as SEO's younger, weirder cousin. Instead of tracking where your brand ranks on Google, GEO tools track how your brand shows up when someone asks ChatGPT or Claude a question.

Havas built Brand Insights AI using Claude Code and Replit. It generates prompts based on a client's brand, runs them across multiple AI models, and analyzes how often the brand appears in responses. They rolled it out across nearly 100 countries and 60+ languages. Then they started licensing it to clients as a SaaS product.

Read that again. An ad agency turned a weekend project into a SaaS revenue stream.

The Real Story Isn't the Tech

The interesting part isn't that Claude Code makes coding faster. It's what happens to agency economics when building custom tools costs hours instead of hundreds of thousands of dollars.

Broadhead's reasoning was simple. Why pay for SEMrush and Profound when you can build something tailored to your exact workflow? At Anthropic itself, a growth marketer named Austin Lau went from never opening a terminal to building Figma plugins and automated ad workflows. His team cut ad copy creation from two hours to 15 minutes and increased ad variations by 10x.

No engineering team. No six-month roadmap. No vendor negotiations.

The Catch (Because There's Always a Catch)

Not everyone's sold. Some analysts point out that the demo is the easy part. GEO measurement standards don't exist yet. The AI models themselves change fast enough that tools built today might need a full rebuild in six months.

Havas hasn't signed an exclusive enterprise deal with Anthropic. Those contracts can run into "multiple millions" annually, and committing to thousands of licenses risks paying for capacity you don't use. Smart move. Stay flexible when the ground is still shifting.

What This Means If You Run an Agency

Three things to pay attention to.

First, the cost of experimentation just collapsed. If you've been waiting for the "right time" to build internal AI tools, you've already waited too long. Your competitors are shipping tools in evenings.

Second, GEO is real and moving fast. Your clients' brands are showing up (or not showing up) in AI-generated answers right now. If you can't measure that, someone else will sell that service to your client.

Third, the line between agency and software company is blurring. Havas is licensing its GEO tool. That's not agency work. That's product revenue. If your agency can build tools that solve real problems, you're sitting on a business model most agencies haven't even considered.

The agencies that move first won't just keep their clients. They'll take yours.

-- Mark Garza, Laimen AI

If you're running an agency and want to talk about what's actually possible with AI tooling right now, let's connect.