New task. Figure out how to implement, run, and analyze Google AdWords campaigns.

My first thought? Damnit, I’m going to have to figure this out on my own.

Or so I thought.

I’ve been using AI heavily for a while now — web scraping, writing, translation, idea generation. I’m comfortable with ChatGPT and pretty fluent in prompting.

But this time I got curious:

How much more could I actually delegate to AI?

That’s when I discovered MCP.

The Model Context Protocol (MCP) is an open-source standard introduced by Anthropic in 2024 that allows AI models and agents to seamlessly connect with external data sources, tools, and software systems.

For me, this meant I could create my own AdWords expert.

Not just a chatbot that explains concepts — but an agent that could connect directly to my Google Ads account, make sure campaigns were structured according to best practices, analyze performance data, and provide structured recommendations.

I started tinkering.

Shout out to this video by Grace Leung that helped me get started.

Setting up Claude in my terminal was straightforward.

Connecting it to AdWords (and YouTube) felt intimidating.

What questions do I even ask? I don’t know half of these terms. What does “good” look like?

What worked surprisingly well was opening two windows — Claude in my terminal and Claude in the web app.

I would get setup details in the terminal, take screenshots, paste them into the web app, and ask: “Does this look OK?”

It was back and forth. Iteration after iteration.

The ability for AI tools to examine screenshots and give feedback is a game changer — and I suspect most people still aren’t taking advantage of it.

Side note.

Today at Home Depot, I helped a couple find the exact light bulb replacement they needed.

I took a photo of the bulb they brought from home, snapped a picture of the shelf options, and asked ChatGPT which one was the right alternative.

Voila.

Everyday augmented intelligence.

After some iteration, AdWords was connected and my new AI data analyst was live.

What would have taken me weeks took two days.

Insane.

And I’m not alone in experiencing this kind of acceleration.

Obie Fernandez recently shared something that stuck with me:

Enterprise software vendors charge six and seven figures for knowledge management systems. They’re built by teams of dozens over years. They require consultants to implement and months to configure.

I built equivalent functionality in four days. While doing my day job as a CTO. Yes, I’m exceptional, but so what? I can only do it today because the tools have changed so dramatically.

There is so much happening right now in AI, and the implications are still unclear.

But from my experience, one thing feels increasingly likely:

Roles will consolidate.

Instead of having one AdWords Manager, one Designer, and one Product Manager, companies may hire one highly capable operator using AI to do all three.

I know this poses real threats to traditional job structures, and the implications deserve serious thought.

But like most people in my space, I’m here for the ride.