#17: Why agencies are starting to make sense again

Hey friends, and a special welcome to new subscribers!

(feel free to connect with me on LinkedIn and say hi! It’s always fun to chat with new readers.)

Key takeaway today: If you run an agency, it’s time to lean into AI. If you don’t run an agency, here’s why you should consider starting one from an AI-first POV:

Let’s start with a trip down memory lane.

Early in my career I was a web designer at an agency. It was actually pretty fun—until I figured out the math.

Selling time is (in most cases) not a great business model. Want more projects? Hire more people. Costs go up linearly with revenue. No leverage, no margin improvement. Just more headcount and overhead.

So I spent the next 15 years building SaaS products instead. Build once, sell repeatedly. Margins improve with scale.

I wrote off agencies completely. Until now.

The Thing That's Different This Time

I'm not about to tell you "AI changes everything"—but it does change agency economics in ways that actually matter.

You can now build real leverage into service businesses. Two ways:

• Multiply bandwidth without multiplying payroll. Build AI workflows that handle routine work instead of hiring more designers. Your bottleneck becomes strategy and oversight, not execution hours.
  

• Price on outcomes instead of time. When AI handles the grunt work, you're selling expertise and results, not logged hours. Big difference in pricing power and margins.

This gets you closer to product-level economics than traditional agency math ever could.

The Hidden Upside

Agencies are incredible discovery engines. You see real demand, real pain points, real willingness to pay. Use that intelligence (and cashflow) to build the product clients keep asking for.

Services → demand → product.

We launched Mojo Works a few months back—an entertainment innovation studio. My first foray back into agencies, built on this exact foundation. Everything approached AI-first.

The window feels wide right now. Most agencies still bill hours for work AI could compress by several x's.

If you run an agency, I'd bet the farm on this transition.

If you don't, picking a niche where competitors are trapped in time-for-money thinking could be interesting.

Tactic: The Resource Library

The best way to increase AI leverage isn't building some do-all superagent. Not yet.

Instead, find small friction points and solve them one by one.

Here's one of my most useful workflows:

I collect resources constantly—tweets, videos, articles—for work and learning. My old process was broken:

1. Find something online

2. Add to todo list

3. Process links a few times per week

Doesn’t sound so complicating? But, when there’s a lot of content. Time adds up.

Too much meta work.

Now I use this workflow:

1. Quick capture: Add resource to my Notion library via phone shortcut
   

2. Auto-process: This triggers classification and processing based on type (summarize YouTube videos, scrape LinkedIn posts, etc.). I use Relay for this.
   

   

   
   
   

3. Clean inbox: Processed items show as "Unread" in my dashboard

   

It’s not the wildest of automations, but every time it runs it saves me a minute or two. And it makes me smile.

The best way to up your AI automation game is by finding small tasks like this one. Build from there.

If you have ideas you want me to write about, hit reply!

Until next week,

Martin