I've been thinking a lot lately about what AI means for experienced operators, agency owners and executives who have spent 15, 20 or 30 years getting really good at something.
Most of the conversation around AI and startups seems to start with the technology. What can we build now? What can AI automate? What business could I start using AI?
I think that's backwards.
For most of my career, there was a pretty significant barrier between seeing a business opportunity and actually being able to build something around it. You could understand a market incredibly well and see a problem that needed solving, but turning that insight into software meant finding developers, raising capital or writing a pretty substantial check yourself. Then you waited months—sometimes a year or more—to find out whether the thing actually worked.
A lot of potentially good ideas died right there. The person who understood the problem wasn't necessarily the person who knew how to build the solution.
AI is starting to change that equation in a pretty profound way.
A nontechnical founder can now get much closer to a working product before raising money. Small teams can prototype things in weeks that once required much larger development efforts. And we're quickly reaching a point where the ability to build something isn't nearly as scarce as it used to be.
Which means the valuable question is shifting from “Can we build it?” to “What should we build?”
That's where I think experienced operators have a really interesting advantage.
Knowing an industry is different from knowing about an industry
AI can help anyone research an industry. It can summarize the competitors, analyze a market, find trends and tell you how companies generally make money.
That's useful. I use it all the time.
But that's different from having spent 20 years inside the business.
If you've been around an industry long enough, you know which problems sound important but really aren't. You know which terrible processes people complain about constantly but aren't painful enough to pay to replace. You know where the budget actually sits, who really makes the buying decision and which “inefficiencies” exist for reasons an outsider won't understand.
You also know where the bodies are buried.
That context is difficult to get from a market report because a lot of it isn't written down anywhere. It's learned by doing the work over and over again.
And I think that kind of earned context is becoming more valuable as the cost of actually building technology comes down.
Don't start with AI. Start with the thing that keeps happening.
If somebody came to me today and said, “I want to build an AI company,” my first question probably wouldn't have anything to do with AI.
I'd want to know what they've seen.
What problem have you encountered 30 times?
What's the spreadsheet every company in your industry has quietly built because the existing software doesn't quite do what they need?
What's the process everybody hates but has simply learned to live with?
What information lives inside the head of one experienced employee who everyone calls when something unusual happens?
What are customers repeatedly paying smart people to figure out manually?
Those are much more interesting questions to me than, “What could we do with AI?”
Because AI isn't the business. It's potentially a new way to solve the problem.
And sometimes the best problems don't look particularly innovative. They look boring.
It might be a reimbursement process, a compliance workflow, an awful handoff between departments or something somebody has been tracking in Excel for the last 12 years. If you've been inside the industry long enough, you've probably stopped noticing how ridiculous some of these things are because they're simply “how we do it.”
Whenever I hear that phrase, my ears perk up a little.
“That's just how everybody does it.” There might be something there.
The things you know probably don't feel valuable to you anymore
This is something I've noticed with experienced people in general.
Once you've been doing something for a long time, your own judgment starts to feel obvious.
You can walk into a situation and quickly recognize that a client is going to become a problem, a project was priced incorrectly, a particular hire probably isn't going to work or a sales process is breaking in a very specific place.
You don't necessarily think, Look at all this intellectual property I possess.
You just think, I've seen this movie before.
But that's exactly the point.
A lot of what we call instinct is really pattern recognition accumulated over years. You've seen enough situations that you recognize things earlier than somebody without that experience.
AI creates an interesting question around that:
Can some portion of that pattern recognition move out of your head and into a product, process or company?
Not all of it can, and I don't buy the argument that AI is suddenly going to replicate everything an experienced person knows.
But some of it absolutely can.
And that's where your expertise starts becoming something more than your hourly value.
So how do you know if there's actually a company there?
This is the part where I think you have to be careful, because not everything you know should become a startup.
There are a few questions I'd want to answer before getting too excited about building anything.
Does the problem repeat?
If every customer has a completely different situation that requires completely different judgment, you may have the makings of a very good consulting business. That's not a bad outcome. But companies become particularly interesting when you begin seeing the same underlying problem over and over.
Is somebody already spending money or significant time trying to solve it?
I'm much more interested in existing pain than trying to manufacture demand. Maybe companies are already paying consultants, hiring people, buying inadequate software or maintaining some awful manual workaround. If money and resources are already flowing toward the problem, that's a useful signal.
Do you know something about the problem that an outsider probably wouldn't?
Maybe the person everybody thinks is the customer isn't actually the buyer. Maybe the important data doesn't live where people assume it does. Maybe the real problem happens three steps before the obvious one.
That's the kind of thing 20 years in an industry teaches you.
Has technology changed the economics of solving it?
This is where AI becomes interesting.
Could something that previously required ten hours of senior-level work now require two? Could a process that was too expensive to automate suddenly become economical? Could one person's expertise be applied across hundreds or thousands of customers instead of dozens?
Now we're getting somewhere.
Can the value eventually exist without you personally delivering it?
If every customer still requires you, you've created another job for yourself. It might be a better job and it might be a very profitable one, but that's different from building an asset.
At some point, something has to move from you into the company—the method, the workflow, the technology, the data, the brand, the people or, ideally, some combination of them.
That's the transition from expertise to equity.
Service businesses may actually have an unfair advantage
Agency owners, consultants and fractional executives are in an especially interesting position right now because they have something a lot of startup founders spend years trying to acquire: proximity to problems.
Your clients tell you what's broken.
Better yet, they pay you to fix it.
You get to look inside multiple companies and see how each one approaches roughly the same challenge. Over time, patterns emerge.
Most agency owners understandably look at client work and ask, “How do we get more of this?”
There's another question worth asking:
“What is all of this client work teaching us should exist?”
What are you building manually over and over?
What analysis do clients repeatedly ask you to perform?
What have you created internally because none of the available tools are quite good enough?
What knowledge lives with your best employees that customers consistently value?
I'm definitely not suggesting that every agency should suddenly become a software company. I've been around enough shiny-object distractions to know how badly that can go.
I'm suggesting you should notice the patterns.
Your service business may be doing more than generating revenue. It may be giving you incredibly valuable market research about the company you could build next.
AI makes bad ideas easier to build, too
There's a flip side to lowering the cost of building things.
It's also easier than ever to build something nobody wants.
You can now spend six weeks building the wrong product instead of spending 18 months building the wrong product. I suppose that's progress, but it doesn't change the underlying problem.
The fundamentals still matter.
Talk to customers. Understand how they solve the problem today. Figure out who actually buys, where the budget comes from and whether the pain is significant enough for somebody to change their behavior.
AI can dramatically compress the distance between an idea and a testable product.
It shouldn't eliminate the testing.
In fact, because building is getting easier, I think the discipline to not build yet may become even more important.
There's another advantage experienced operators have
Distribution.
This gets dramatically less attention than product, but it's often just as important.
If you've spent two decades in an industry, you may already know your first 50 potential customers. You know the conferences they attend, the associations they belong to, the people they listen to and how purchasing decisions actually get made.
You may have credibility with people a first-time founder would spend years trying to reach.
That's part of your accumulated experience too.
When you combine deep understanding of the problem, a new ability to build relatively inexpensively and existing access to the market, the profile of who can realistically start a technology company begins to change.
For a long time, the startup mythology has been dominated by the young technical founder with a breakthrough idea.
That person isn't going anywhere.
But I think we're going to see more 40-, 50- and 60-year-old founders who come at this from the opposite direction. They aren't starting with the technology. They're starting with decades of accumulated knowledge about a problem and realizing that technology has finally made it practical to build the solution they always wished existed.
That's interesting to me.
Your experience doesn't automatically make you a founder
There's one important caveat.
Being an expert in an industry doesn't mean you automatically know how to build a company.
You still have to validate the idea, position it, price it, sell it, build the product, create distribution, hire people and make a thousand decisions that may have nothing to do with the career you've had up to this point.
Deep expertise gives you a potentially powerful starting point. It doesn't give you a free pass.
In fact, one of the hardest parts for an experienced operator may be becoming a beginner again.
You can be 25 years into a successful career and suddenly find yourself asking basic questions about product architecture, venture economics, fundraising or something else you've never had to understand.
That's okay.
Bring the thing only you know.
Then surround it with the people and capabilities required to turn that knowledge into a company. And when a growing company needs someone to own the work—not just advise from outside—understand the distinction between a consultant and an embedded operator.
Maybe the company has been hiding in plain sight
If you've spent 15, 20 or 30 years becoming really good at something, I wouldn't start by asking:
“What AI company should I build?”
I'd ask:
“What do I understand today that I couldn't possibly have understood 20 years ago?”
Then I'd look at the problems that knowledge allows you to see differently.
And then I'd ask whether there's a way to build something around one of those problems that creates value without requiring you to personally deliver the answer every single time.
That's a very different starting point.
AI may make the product easier to build, but your job is still to find the thing worth building.
And there's a decent chance you've already spent your career learning where it is.
That possibility is at the center of Buildside's The Hidden Company idea.
Is there a company hiding inside what you already know?
Buildside works with experienced operators, agency leaders and founders to identify the opportunity, validate the problem and build the business around it—from the first idea through go-to-market and operating infrastructure.
Start a conversation ↗Brian Cross is a founder, operator, investor and embedded executive who has spent 25+ years building companies from inside the work.