AI in 2025: Love It or Leave It.

In the post-AI world, either integrate with artificial intelligence or skip it altogether.

Written by Joe Procopio
Published on Dec. 06, 2024
A big robot is holding a laptop.
Image: Shutterstock / Built In
Brand Studio Logo

I wouldn’t want to be the founder of a strictly tech company right now.

I’ve been slinging cutting edge technology for going on 30 years — with about 15 of that in generative and traditional AI – sometimes as a founder of my own company, other times riding along with brilliant people doing amazing things while risking large sums of venture capital to produce outsized returns.

4 Ways Tech Founders Can Tackle AI

  1. Go for a niche approach.
  2. Create a killer use case.
  3. Promote the human element of your business.
  4. Bring back real customer support.

But I’ve never seen the tech ecosystem struggle like this. The environment is ugly, weak, scary — pick your own depressing adjective.

Why? AI. It’s the elephant in the room or the 800-pound gorilla, standing there, staring at you. Don’t make any sudden moves.

We can probably agree that AI is here to stay and we’re now living in the slightly dystopian remains of a post-AI world, version 1.0 anyway. In fact, as I write this, a lot of talk is swirling around the beginning of the AI Wall.

If I was a tech founder now — and let’s face it, it’s just a matter of time before I dive back in— I know for sure that I have to at least acknowledge the AI elephant-gorilla, if not incorporate it into my plans.

So after all the talk and hand-wringing with my peers, I came up with what I believe are the four best ways to tackle AI head-on as a tech founder, regardless of how much AI is already in your current plans.

Further ReadingWhat Is Artificial Intelligence (AI)?

 

Plan A: Go Niche

Like any good Plan A, this one is about going all-in, in this case, all-in on AI while acknowledging the fact that the Version 1 game is over and the winners have been selected. And unless you have a war chest with about $1 billion of VC and corporate investment behind you, well, I hate to be the messenger here, but you’re not one of the winners this time.

So what do you do?

The thesis for my AI Wall article is that the wall is about more than tech and runs deeper than processing limits. It’s more an existential threat of the law of diminishing returns up against the model of the current version of AI trying to be all things to all people

In other words, we’ve reached the moment where no matter how much data we throw into the learning model, the results just aren’t going to get much better for you, me, or society as a whole. We’ve reached maximum general productivity at minimum cost. This happens with every massive new wave of technology.

The more you can niche your model while maintaining the broadest audience, the better your chances for success. For now.

When it happens, the path for folks like you and me — and in fact it’s my belief that this should be the only path with AI — is to serve a niche instead of a broad spectrum. It’s what I was doing at Automated Insights back in 2010 when we first got on this generative AI train. In today’s generative AI, we’re starting to see this with RAG techniques being applied to LLMs, turning the general use cases, like what you see in those Google Gemini ads, into more niche services.

AI agents for accounting. AI agents for construction. AI agents for parenting. And so on. 

It’s like a sliding scale. The more you can niche your model while maintaining the broadest audience, the better your chances for success. For now. 

 

Plan B: Go Killer Use Case

I’ve spoken about this dozens of times and I did again in my AI Wall article. Again, back in 2010 when I was at Automated Insights, we learned early on that our use case — machines writing narrative articles from data — needed to be modified a bit, into machines writing narrative articles from data where humans could not.

Once we hit on that killer use case, people started paying us to fill that use case. 

In my somewhat tongue-in-cheek article about Daisy The AI Granny, a platform from UK mobile giant O2 that purports to slow or stop phone scammers, my real intention was to show the inherent value of this particular use case. I contend that while the use case might not drive much revenue for O2, a small tweak to the use case is likely AI success waiting to happen. 

In today’s AI business environment, we’re seeing a lot of AI solutions in search of problems. That thinking needs to be flipped on its head. If you’re working on solving a big, hairy, expensive problem, ask how can AI lend itself to that solution? Or can it at all? 

And if the answer to that question is “it really can’t,” then instead of trying to shove a square AI peg into a round problem hole, zig while everyone is zagging and move on to Plan C.

 

Plan C: Go Human

Recently, I’ve adopted a selling point with my writing, touting the fact that my posts are straight, no-chaser, honest, human writing in a sea of flat, soulless, generative AI nonsense and misinformation.

That really resonates.

Now, writing is one thing. How does this apply to business, especially a technology business?

Well, the hurdle to get over with this plan is that the possibilities are endless but the right answer is unique to your business, hard to quantify, and may not even exist.

Yeah, sorry. I didn’t say I had solutions, I said I had plans.

But let me ask you this. Are you sick of AI yet?

Well, so is everyone else. Much like you would ask, “How can AI lend itself to my solution?” you might ask yourself the opposite, “How would a human touch help sell my solution against AI exhaustion?”

 

Plan D: Go Full Support

Positioning your offering against AI could be as simple as offering live 24/7 human support as an alternative to a chat box that, let’s face it, most humans realize is just going to spit back statements from the company FAQ. 

In fact, there’s a greater case to be made here, one about not running your business like a business at all. This is a longer thought that needs its own post, but the basic message is this:

As we’ve been approaching AI, through digital automation, internet and mobile reach, and the generative AI dabbling that folks like me started doing a decade ago, the entire way corporations conduct business with customers has evolved into the exact opposite of what corporations should do to show customers they value their business. 

This manifests itself in everything about a company, product or service that looks like it was a decision made to support the company at the expense of the customer, and it extends to pricing and usage models, prioritization of convenience, reduction of consumer choice, and especially weak support and limited contact. 

AI is the top of that tower of garbage.

Ultimately, paying for a dedicated human support team might end up being a cheaper overall solution than paying for hack AI delivered via chat that just makes customer problems worse. And when you add in the dividends of being able to sell against the competition’s dirt-poor customer experience, it might even be a profitable move.

More From Joe Procopio4 Massive Minimum Viable Product Mistakes You’re Inevitably Going to Make


These Are Good Times for the Right Ideas

Philosophically, at least, some of the best and most successful startups emerge during these kinds of periods when everyone is zigging and they zag. If there’s a single ray of sunshine among all these clouds, it’s coming from the true maverick entrepreneurs who are either finding inefficiencies to exploit or attacking the AI hype head-on. 

Both Automated Insights and my own solo-founded startup ExitEvent began life at the tail end of the Great Recession and had nothing to do with mobile app tech, which at that time was sucking all the oxygen out of the room for tech founders like AI is today. 

If you’re looking to make a buck off the AI gravy train, you’re probably too late. But if you’re working on something truly innovative that needs incubation and funding, well, the risk is going to be just as high as it ever was, but the reward is potentially just as big if not bigger.

Explore Job Matches.