I've Seen This Movie Before: Surviving the AI Gold Rush

In 1999, eToys — an online toy retailer barely two years old — carried a higher market cap than Toys "R" Us. Read that again. A startup was worth more on paper than the company whose shelves held more toys than any retailer in the history of retail.

So many experts said "shopping is going to move online in an instant", and they were bold enough to almost state the day and time that it was going to occur. And not only did the financial markets believe them, it poured money into them. By 2001, eToys was bankrupt, along with most of the other dot-com startups that had equally crazy valuations.

I didn't read about this in a case study. I lived and worked through it, along with many other tech cycles in my career. Twenty-five years later here we are again - the AI gold rush is running the same script and telling the same stories, with the newest of new technologies.

What follows below is the lens that I use to filter out the real opportunities in the gold-rush stories — built through years of living through the adoption of countless new technologies. Even if pick up nothing more from this post, I hope you pick up that lens.

Is the AI Gold Rush Just the Dot-Com Bubble Again?

In pattern, hype and threat, yes. Every tech gold rush produces the same two bets: the loud bet that makes the headlines, that everything changes overnight and - and the quiet bet that adoption happens more gradually one pain point at a time. The real money, the sustainable money goes to whoever solves the real problems that people already live with every day...the real pain of life and business.

The real, sustainable adoption of any given technology happens when that real pain of a given problem exceeds the friction or the risk of the new tool or technology.

That is the lens, the "pain formula", and the rest of this story is what it looks like in the wild.

The Bet That Became Wayfair

While eToys (and many others) rode the overnight-transformation story, two guys from Boston made the other bet. Niraj Shah and Steve Conine founded CSN Stores in August 2002 — a two-person operation run out of Conine's nursery. Their first site wasn't a marketplace for everything. It was racksandstands.com. Entertainment furniture. That's it. Guess what, I bought from them – they had the best selection anywhere at the time.

Their insight was pure human nature: people would first buy online the things they couldn't find in retail — because the pain of not having the item had to exceed the fear of typing a credit card number into a "black box" (aka, a website). In 2002, that fear was real. So they went where the pain was highest, one unglamorous niche at a time: everycuckooclock.com, simplydogbeds.com, thewokstore.com, justsouthwesternrugs.com — until the roster passed 200 niche eCommerce stores. Each one focused. Each one solving a single unfindable-item problem. In 2011 they consolidated the whole catalog into one brand: Wayfair. They went public in 2014.

The experts of 1999 were directionally right — shopping did move online, en masse. They were catastrophically wrong about how and when. It didn't switch overnight; it moved one pain point at a time. eToys had the story. CSN Stores had the formula. That's the entire CSN Stores / Wayfair story in one line, and it's worth more than any forecast you'll read this year.

I Wasn't Watching The Movie — I Was in It

My connection to this isn't academic. In the early '90s video game era, I was selling into the very big-box shelves eToys was supposedly "going to kill." I watched the gap between what analysts predicted and what customers actually did — from inside the aisle.

Then I spent fifteen-plus years running niche/micro-niche eCommerce stores myself: LadysGolf.com, StyliSource.com, StylusCentral.com, e-Tether.com, and eventually a whole catalog of stores built on one architecture designed to clone and deploy new ones fast. The same species as the CSN model, lived on a bootstrapped scale. Those years taught me two things the case studies leave out: the niche-catalog model works — and it demands two things underneath it, the right technology shift and real fuel behind it. You can be right about the model and but still need to find the right opportunity to capitalize on it.

The "Déjà Vu All Over Again" at C10 Labs

Two years ago I was mentoring at C10 Labs in Boston, and I started hearing the same prognostications: AI transforming the world overnight. Big platform plays. Breathless valuations. Everything, instantly. Sound familiar?

Meanwhile, my friends running small businesses were suffering the same old operational problems — and ChatGPT could only chip away at them on an ad hoc basis. Helpful for big problems on a small scale. Nothing systematized for larger scale. The distance between the pitch decks and my friends' actual Tuesdays was the same distance I'd seen between the 1999 analyst reports and the retail shelf.

I kept saying it over and over again: I've seen this movie before!

The Proof Moment

Amid this timeline, I was working to help my best friends get ScandinavianShoppe.com back on the growth trajectory they were striving towards. Among other things required efforts was moving their SEO metrics (and site traffic) in the right direction. The problem was that so much of their long-time success in SEO was rooted in the original product descriptions and meta data they created from 2010-2020. It all needed to be refreshed and rewritten...7,000 products worth.

Monumental undertaking – absolutely, but it was 2024...the "Age of AI" upon us...so I started experimenting with ChatGPT to do the rewrites, optimize the meta data, etc., making sure best practices were employed. Once we got the prompting right it worked beautifully! Then I did the math...between copying and pasting, generating and proofing the new descriptions and the meta data, it was going to take hundreds of hours to cover their catalog. Worth it – yes, practical – no!

Surrounded by so many geniuses at C10 Labs I asked one of them if the formula we created could be scripted. While not without the bumps and bruises associated with any new technology we got it done, 7,000 products refreshed. The impact to ScandinavianShoppe's SEO was like none I had every seen before...a 25% keyword ranking boost...and a new product was born.

That was the moment the threads fused. The gap between "AI can do this" and "AI does this, at scale, inside a real workflow" is the "last mile of AI" — and the last mile is where the money is. I'm not the only one who landed there. Faisal Feroz, in an essay called The Bottleneck Economy, puts it plainly: that last mile "is where most of the money will be made," and "problem finding eclipses problem solving." From where I sit, that's exactly right. Hands don't scale; systems do. The AI gold rush is overflowing with demos. The value sits in the unglamorous work of carrying a proven capability that last mile into a business's actual catalog, actual inventory, actual Tuesdays. That is what practical AI for most businesses looks like: not transformation, but a specific pain point, removed, at scale.

How to Use This Lens

The next time a gold rush claim, pitch, or purchase crosses your desk for AI or any technology — whether you run a shop, a sales team, or a startup — ask three questions:

1. Does this alleviate a pain we feel NOW?

Not "could this help someday?" — is someone losing hours, orders, or sleep to this problem today? CSN didn't create demand for hard-to-find furniture; the pain existed already. So for the technology / solution you're looking at, is it focused on a problem that we haven't had the time, resources or ability to solve???

2. Is the pain of the problem you are trying to solve greater than the pain of implementing the solution (time, money, learning curve, etc.)?

Every new behavior has a cost: learning it, trusting it, changing a routine around it. Adoption happens only where the pain of the problem exceeds that "friction". If the tool demands a new workflow to solve a mild annoyance, it will be piloted and abandoned. Conversely, if it's easy to deploy and solves a major headache it will be adopted quickly and in use for a long time.

3. Who carries the last mile — and do they have anything left?

Every fix lands on somebody. In a small business that somebody is usually already carrying nine other things. So the question isn't whether the fix is a good idea. It's what the costs is for the person who has to make it work, and what they need to stop doing in order to focus on it.

Often the honest answer is that it just doesn't get done. The SEO doesn't get done. The physical inventory never gets reconciled. Everyone agrees it matters; nobody has the hours. (Friction Hunting is the long version of that exercise.)

Make that list — the things that only happen if YOU do them, and the things that would move the business if anyone had time to implement them. That list is your ceiling. It's what keeps a $2.5 million company from becoming a $5 million company, and a $5 million company from becoming a $10 million one. Not strategy. Not the market. Just the person who carries the last mile running out of gas.

The tools worth buying are the ones that take something off that list without adding a new job to it.

Anything that fails these questions is gold-rush story, a good story maybe, not business. Anything that passes them will quietly compound while the headlines chase transformation.

The Other Bet, Again

One more parallel sealed this model for me. In 1982, Trip Hawkins founded Electronic Arts — shaped like United Artists, the 1919 studio "built by the stars." EA didn't build its own games in those early years — every title came from outside — and that is exactly how it won, as a publisher: giving independent creators' niche titles a platform to be built, marketed, and scaled. CSN proved the niche-catalog model in stores; EA proved it in software. Different payload, same rocket.

So that's the bet I'd personally make: a growing catalog of small, focused AI applications, each one killing a specific, costly friction in a real business workflow — built and published the way CSN built stores and EA published games. Not because catalogs are fashionable, but because I've seen this movie before, and I've seen which characters tend to be around when the credits roll.

The AI gold rush will do what gold rushes do. The pain formula will do what it has always done. Place your bets accordingly.