Thoughts on Scaling in the Age of AI

Fri, Apr 12, 2024

We are all obsessed with scale. You, me, and every entrepreneur I have ever known. 

In fact, a near bulletproof way of dismissing any idea is to say, “It doesn’t scale.” Historically, the ability to scale has been a sound measuring stick. That’s not true anymore and I’ll explain why. 

Take anything that has the innate ability to scale and you will have identified something that AI either: a) can do right now, or b) will be able to do very, very soon.

If it scales, AI can do it. This is the new rule of entrepreneurship everyone needs to come to terms with yesterday morning. 

If AI can do it, its value will be nullified down to the commoditized price of the computing power required to fulfill whatever process we’re discussing. Any margin that exists for profitability will need to be created. Here’s where this conversation gets fun and a little cyclical. 

If margin can be created by adding additional processes, and those processes scale (e.g., AI can perform those processes), that margin instantly gets flattened according to the same rate of measure. 

Meaning: Only processes that don’t scale will yield adequate margins for profit.

Here’s a made up example: We have a business where we take every question that’s ever been answered via support ticket systems like ZenDesk and turn them into social content. Pre-AI, this was an amazing engine with massive scale potential that all relied on human-driven processes.

Post-AI, this is a day’s worth of work sewing together ZenDesk and ChatGPT via Zapier: Support ticket + LLM + custom prompts = scalable output. Our business is doomed. What used to be an entire agency is now just a one-button feature waiting for ZenDesk to develop natively.

So, how do we save ourselves? What we need to start doing is looking for the things that don’t scale. 

Those are the places where we’ll be able to plant ourselves and create the necessary value (i.e., margin) to justify our existence and keep from being supplanted.

In the case of this made-up business, maybe it’s taking the content that’s created and deciding when, where, and how it should be distributed based on the company’s promotional schedule. Or, simply choosing which posts to schedule for repurposing based on the company’s goals.  

The most obvious of the non-scalable opportunities are in expert-led, human-to-human connections. 

People are still going to want to talk to people. AI might be a more effective diagnostician than any doctor, but who wants to find out they need surgery from a computer? 

While that example is a little morbid, it’s a pretty solid illustration of how we need to start thinking. That’s a conversation that we can’t (or, at least, shouldn’t) attempt to scale. That makes it a perfect place to niche down as a provider and add value AI can’t replicate. 

Your turn: What are the safe harbors in your business? What areas can’t and won’t scale?