I Recorded a Voice Memo While Brushing My Teeth. By Morning, I Had 12 Live Websites.

The Idea That Usually Dies

A couple of weeks ago I was brushing my teeth and had one of those random ideas. The kind you'd usually forget by the time you finish flossing.

I run an adult party card game called FTG. It uses the Stroop Effect, so it's part brain training, part swearing, and it's not safe for work. I was thinking about traffic, and the idea popped in. A Shakespearean insult generator. Quick, fun, tied loosely to the card game vibe.

In the old world that idea dies right there. Too random to brief to a designer. Too small to pay an agency to build. Too weird to test.

But I've got an AI agent system running 24/7 across my businesses now. So instead of forgetting it, I pulled out my phone and recorded a 50 second voice memo.

Then I went to bed.

By morning, 12 live websites.

The Idea Got Better Overnight

This is the part most people miss when they talk about AI building things. Building isn't the hard part anymore. Making the right thing is.

When my voice memo hit the system, it didn't just go "cool, here's your insult generator." It pushed back. It knows my business. It has context on the most recent sales, the recent marketing activity, what's working, what isn't. So it said something like, "Hey, the insult generator is a fun angle, but if the goal is traffic to the FTG store, we should also test quiz and trivia formats, social and party game tools, historical themes, and a few others."

So instead of one site, it scoped 12. All free little web apps. All deeply tied to what FTG actually is, so the audience overlap is real. All with copy, SEO, generative engine optimisation for AI search, sound effects, the lot.

That's the thing I keep coming back to. The system made my idea better before it touched a line of code.

How the AI Marketing Team Actually Works

I think of these agents as virtual staff. Each business I run has its own org chart.

For FTG, the request lands with a Chief Marketing Officer agent. The CMO doesn't do everything itself. It delegates, the same way a real CMO would.

The Research Layer

The CMO spins up subagents to research everything relevant. What people search for. What competitors are doing. What angles haven't been touched. The research is wide because the goal is wide — drive traffic — so it casts a big net.

The Specialists

Then the specialists weigh in.

An SEO strategist handles search and GEO (generative engine optimisation), which is how you get found by AI agents and AI search now. That's the new battleground and most businesses aren't even thinking about it yet.

A brand voice agent has all the documentation on FTG, the tone, the product, what we sound like. It makes sure every site reads like us, not like generic AI slop.

A sound designer researches what audio cues belong on a party game app. What does a wrong answer sound like? What feels rewarding?

A copywriter turns all of it into actual page copy.

A compliance checker matters a lot for FTG, because it's adult content. Things have to be flagged correctly. The compliance agent makes sure we're not getting ourselves into trouble.

The Build Team

Once the strategy is locked, the CMO writes a plan and hands it to the dev side. A lead developer takes the plan, delegates the actual building to coding agents, and a QA agent checks everything works on the web. By the time I see it, it's built, tested, and live.

All of this happens while I sleep.

The Result: 12 Sites, 3,000 Views, Zero Ad Spend

One week in, those 12 sites are pulling around 3,000 views. No ads. No paid promotion. Just organic traffic from people searching for the kinds of things the system identified.

Every one of those sites links back to the FTG store. So it's not just vanity traffic — it's traffic with a path to a sale built in.

A week is early. I'll come back in a month or two with the real numbers. But the direction is clear, and the cost to get here was a voice memo I recorded with toothpaste in my mouth.

Why This Matters If You Run a Business

The thing that's actually changed isn't "AI can build a website." That's been true for a while. What's changed is the cost of trying an idea.

12 little sites like these, scoped properly, written properly, built properly, would have cost me $10K minimum from an agency. Probably more. Definitely weeks of back and forth. And because of that cost, I'd never have done it. The idea was too speculative. Too "let's see what happens."

That's the trap most small business owners are stuck in. You only test ideas you're already confident in, because testing is expensive. Which means you never test the weird ones. Which means you miss the ones that would have worked.

When testing is basically free, and the system makes your idea better before it builds it, the game changes. You try the random stuff. The Shakespearean insult generator. The thing you thought of brushing your teeth.

Some of it works. Some of it doesn't. The cost of finding out is a voice memo.

Want One of These For Your Business?

This is what we build at AI Workshift. AI business partners that know your business, push back on bad ideas, and execute the good ones while you sleep. Not chatbots. Not generic automation. Actual virtual staff with real context on what you're doing and why.

If that sounds useful, get in touch or just DM me. Happy to walk you through what a setup like this would look like for your business.

I'll report back on how the 12 sites perform.