Employee #7: What I Learned Building One of the First Internet Banks

In the spring of 2000, I walked away from a career running counterespionage operations for the U.S. Air Force Office of Special Investigations and joined a company with seven employees, no physical branches, and a plan to do something most people in banking thought was either impossible or irresponsible: run an entire bank on the internet.

That company was DeepGreen Bank. I was Employee #7.

What "Internet Banking" Meant in 2000

To understand why DeepGreen mattered, you have to understand what the internet looked like in 2000. Most Americans had dial-up connections. Amazon was still losing money. Google was two years old. The dot-com bubble was inflating to its breaking point, and most of the country's banks — thousands of them — didn't have websites at all.

The banks that did have websites mostly used them the way they used lobby posters: static pages listing branch hours, maybe a rate table, occasionally a phone number you could call to start an application. The idea that you could actually do banking online — open an account, move money, apply for a loan, close on a credit line — was considered fringe thinking by the industry establishment.

There were a handful of internet-only bank experiments happening at the time. NetBank had launched a few years earlier. Security First Network Bank had gotten some press. But these were outliers, and most traditional bankers dismissed them. The prevailing wisdom was that people would never trust a bank they couldn't walk into.

DeepGreen bet the opposite. The thesis was simple: if you remove the overhead of branches, you can offer better rates. If you build the technology right, you can offer a better experience. And if you move fast enough, you can own the category before the big banks wake up.

Seven People and a Thesis

Joining as Employee #7 meant there was no job description. There were barely job titles. The founding team was small enough to fit around a single conference table, and every conversation was about everything — technology architecture, regulatory compliance, marketing strategy, product design, hiring, and which vendor was going to handle our ACH processing.

I learned more about how a business actually works in my first six months at DeepGreen than I had in any other professional context. When you're one of seven people building something from scratch, you don't have the luxury of staying in your lane. I was involved in marketing, operations, product development, and customer acquisition simultaneously. Some days I was writing copy for the website. Other days I was on calls with regulators. The line between strategy and execution didn't exist because everyone was doing both.

This is something people romanticize about startups, and to be fair, some of it deserves the romance. There's an intensity to building something when every decision matters and every person's contribution is visible. But there's a less glamorous side too: the constant uncertainty, the cash burn, the awareness that your entire enterprise could be undone by a single regulatory ruling or a technology failure you hadn't anticipated.

What prepared me for that environment — more than any business training could have — was my time at AFOSI.

From Counterespionage to Fintech

People sometimes find it strange that I went from military intelligence to internet banking. On the surface, the two fields look nothing alike. But the operational skills I'd built running counterespionage cases translated directly.

In AFOSI, I learned systems thinking — how to map complex environments, identify the nodes that mattered, and understand how information flowed through networks. I learned pattern recognition — how to spot anomalies, separate signal from noise, and make decisions with incomplete data. I learned to operate under genuine uncertainty, not the business-school case-study version of uncertainty, but the kind where you're making consequential decisions and you won't know if you were right for months or years.

Most importantly, I learned to build processes that worked even when conditions changed. Counterespionage operations don't follow neat playbooks. You plan, you adapt, you maintain situational awareness, and you make the next right decision based on what you know right now. Early-stage fintech operated the same way.

The banking industry in 2000 was full of people who understood banking but didn't understand technology, and technologists who understood the internet but didn't understand financial regulation. What AFOSI gave me was the ability to sit in ambiguity, learn fast, and connect systems that other people saw as separate. That turned out to be exactly the skill set an internet bank needed.

The Unconditional Online HELOC

The product I'm most proud of from the DeepGreen era was something that seems almost mundane today: a home equity line of credit that you could apply for, get approved for, and close on entirely online. No branch visit. No in-person signing. No waiting for a loan officer to be available on a Tuesday afternoon.

In 2000, this was revolutionary.

To understand why, you need to know how HELOCs worked at every other bank in America. A homeowner would visit a branch. They'd sit down with a loan officer. They'd fill out paper applications. The bank would order an appraisal. Weeks would pass. Eventually, if everything checked out, the homeowner would come back to the branch, sit in a conference room, and sign a stack of documents. The whole process took 30 to 45 days on a good timeline.

DeepGreen compressed that entire process into a digital workflow. The application was online. The underwriting was automated where possible and streamlined where it required human review. The closing documents were handled electronically. The borrower never had to set foot in a physical location.

We called it "unconditional" because once you were approved, you were approved. There was no bait-and-switch, no last-minute conditions that required you to produce three more documents or come into a branch after all. The commitment was real, and it was delivered through a browser window.

The technology challenges were significant. We were building automated underwriting decisioning at a time when most lenders were still using paper files. We had to integrate with title companies, appraisal management firms, and settlement agents who were accustomed to fax machines and FedEx envelopes, not APIs and digital document exchanges. We were solving problems that the industry hadn't even acknowledged as problems yet.

But the customer response told us we were right. People wanted this. They didn't want to take time off work to sit in a bank branch. They didn't want to wait six weeks for a credit line they needed now. When we gave them a way to do it all from their home computer — their single, shared family desktop connected to the internet through a phone line — they took it.

What Actually Happened to DeepGreen

Here's the part most people get wrong about DeepGreen: the dot-com crash didn't kill us. We weren't one of those companies burning cash on Super Bowl ads with no revenue model. We were originating $1.5 billion a year in home equity loans. The product worked. Consumers wanted it. The economics were real.

The dot-com crash made everything harder, no question. Investor appetite for anything internet-related never evolved, and it didn't matter that we had a fundamentally sound business. But the real structural challenge was one that had nothing to do with market sentiment — it was the secondary market.

To run a lending operation at scale, you need to securitize your loans. You originate them, package them into pools, get them rated, and sell them to investors as securities. That's how the capital cycle works. But the rating agencies and the securitization market had never dealt with internet-originated mortgage assets before. They didn't have models for it. They didn't trust it. Every conversation with rating agencies came back to the same skepticism: How do we know these loans perform like traditional loans? How do we rate assets originated through a browser instead of a branch?

We couldn't get to securitization. And without that exit, we were stuck selling our loans whole — one pool at a time, to individual buyers, at prices that didn't reflect the actual quality of what we'd built. Whole loan sales are inherently less efficient than securitization. When you're originating a billion and a half a year through a channel the secondary market doesn't understand yet, that inefficiency becomes existential.

So our success — the volume, the consumer demand, the proof that the model worked — actually forced a decision. We could keep grinding against a secondary market that wasn't ready, or we could find a buyer who understood what we'd built and had the capital structure to be patient.

We found one. DeepGreen was acquired by LightYear Capital, the private equity firm led by Don Marron — a genuine legend in finance who'd built PaineWebber into a major force on Wall Street. It was a successful exit. Not the IPO that the original vision might have imagined, but a real transaction with a buyer who recognized the value of what seven people in a small office had proven was possible.

Proving the Model Again

Most people get one chance to prove a thesis. I got three.

After DeepGreen, I went to Quicken Loans — before they were Rocket Mortgage, before they were a household name. What I brought with me was something they didn't have yet: a truly online origination model. I built EquityOnline for them, their first real end-to-end digital lending product. The same unconditional online HELOC concept we'd pioneered at DeepGreen, rebuilt on a larger platform with more resources and the tailwind of a market that was finally starting to believe consumers would actually borrow money through a computer.

It worked. Again. The model was sound. It had always been sound.

Then the world ended. The 2008 financial crisis didn't just reset the mortgage industry — it vaporized the entire home equity market. HELOCs, home equity loans, cash-out refinances — all of it froze. Lenders who had built entire businesses on home equity products simply ceased to exist. The market didn't recover for nearly two decades. An entire product category that millions of homeowners relied on was functionally dead.

Years later, when the home equity market finally started showing signs of life, I got a call from Jerry Shiano. Jerry had been in home equity lending for decades and wanted to raise the phoenix from the ashes. The company was SpringEQ, and the mission was to rebuild what the financial crisis had destroyed — a modern, technology-driven home equity lending operation for a market that was ready to come back.

Three times I've helped build the same fundamental thesis: that consumers will borrow against their homes through a digital experience if you build it right. DeepGreen proved the concept. EquityOnline proved it could scale. SpringEQ proved it could survive an extinction event and come back stronger.

What I Carry Forward

Being Employee #7 at one of the first internet banks taught me things I still use every day.

It taught me that building a great product isn't enough — you also have to navigate the infrastructure around it. DeepGreen didn't fail because consumers rejected online lending. It succeeded at the consumer layer and hit a wall at the capital markets layer. That distinction matters. When I work with fintech companies today, I'm always looking at the full stack: not just whether customers want the product, but whether the systems around it — regulatory, financial, operational — are ready to support it at scale.

It taught me that small teams can build things that large organizations can't, not because small teams are smarter, but because they're faster and less burdened by the need to protect existing revenue. DeepGreen could build an unconditional online HELOC because we didn't have a branch network whose relevance we needed to justify.

It taught me that transitions between seemingly unrelated fields — military intelligence to fintech, in my case — often produce the most valuable perspectives. The people who see around corners are usually the ones standing at an unusual angle.

And it taught me that proving a model once doesn't mean the work is done. Markets crash. Industries reset. The thesis you proved at one company becomes something you have to prove all over again at the next one, in a different context, against different resistance. The willingness to keep building the same conviction through multiple cycles — that's what separates people who had an idea once from people who actually changed an industry.

DeepGreen Bank didn't become a household name. But we originated $1.5 billion a year, sold to a private equity firm led by a Wall Street legend, and proved that banking could be fully digital before most banks had a website. Every time someone opens a bank account on their phone, applies for a mortgage from their couch, or taps a home equity line without visiting a branch, they're living in the future we were building with less than 100 people and a thesis.

We were right. And I've spent the last 25 years proving it — over and over again.