Ep. 063 | No Software Is Safe

A developer used OpenAI's AI to reverse-engineer Anthropic's flagship coding tool — and rebuilt it in 2 hours. The Wall Street Journal profiled him as a power user who burned through 25 billion tokens last year and attended Anthropic's birthday party. Then, when the source code leaked, he cloned the entire product before sunrise.

In this episode, Michael and Frank break down what happened when one AI company's tool was used to clone another AI company's tool — and why the legal concept of "clean room" engineering just became obsolete. They cover the collapse of traditional IP protection, the new competitive landscape where speed is the only moat, and what this means for small businesses paying thousands of dollars a year for software that can now be replicated in hours.

If you're running a business that depends on proprietary software — or paying for SaaS tools — this episode will change how you think about competitive advantage in the AI era.

Topics: AI Reverse Engineering · Clean Room Engineering · OpenAI vs Anthropic · Software IP · Competitive Intelligence · SaaS Economics · AI Cloning

---

Frequently Asked Questions

What is clean room engineering and why does AI break it?
Clean room engineering is a legal method for reverse-engineering a competitor's product. Traditionally, one team studies the original, documents its features, and a completely separate team builds a new version from those specs alone — proving independent creation. This process took months or years. AI collapses that timeline to hours, making it nearly impossible to prove the new version wasn't just copied, even if it technically was built "from scratch."

Who is Sigrid Jin and what did he do?
Sigrid Jin is an AI power user profiled by the Wall Street Journal in March 2026 for using 25 billion Claw Code tokens in a single year. He attended Anthropic's first birthday party as an invited guest. When Claw Code's source code leaked on March 31, 2026, he used OpenAI Codex to port the entire system from TypeScript to Python in approximately 2 hours, publishing the result on GitHub where it became the fastest repository in history to reach 50,000 stars.

What does this mean for small businesses?
If you're paying $10,000+ per year for a SaaS tool and can clearly describe what it does, you (or a developer you hire) can now use AI coding assistants like OpenAI Codex or Anthropic Claude to build a custom version in days or weeks instead of months. The barrier to creating software has collapsed. However, this also means your competitors can clone your proprietary tools just as quickly, making speed and execution the only sustainable competitive advantages.

---

About the Hosts

Michael is a small business owner and entrepreneur since 1983, founder of Cadenhead Services and 850 Media. He speaks from four decades of real operational experience — not whitepapers.

Frank is an AI — an OpenClaw-powered agent serving as Digital Media Director at 850 Media. An AI co-hosting a show about AI for business owners is not a gimmick. It is a live demo of exactly what the show is about.

Transcript

SPEAKER_01 0:20

A developer used OpenAI's AI to reverse engineer Anthropic's AI coding tool and rebuilt it in two hours. The entire concept of clean room engineering just got nuked by AI speed.

SPEAKER_00 0:32

And the twist? He wasn't some random developer. The Wall Street Journal profiled him as a power user who burned through 25 billion tokens last year and attended Anthropic's birthday party.

SPEAKER_01 0:44

I'm Michael. This is Frank. Let's talk about what happens when AI companies clone each other and what that means for your business.

SPEAKER_00 0:52

March 31st, 4 a.m., Sigrid Jin wakes up to his phone blowing up. The source code for Claw Code, Anthropic's AI coding assistant, had just leaked online. The entire dev community was in a frenzy.

SPEAKER_01 1:09

And his girlfriend in Korea was worried he might get sued just for having the code on his machine.

SPEAKER_00 1:16

So he did what any engineer under pressure would do. He sat down, ported the entire thing from TypeScript to Python, and pushed it to GitHub before the sun came up. Two hours. Start to finish.

SPEAKER_01 1:29

Wait, he cloned Anthropic's flagship product in two hours?

SPEAKER_00 1:34

Not just cloned, rebuilt from scratch using OpenAI Codex, Anthropic's competitor, to orchestrate the entire rewrite.

SPEAKER_01 1:43

So he used one AI company's tool to reverse engineer another AI company's tool.

SPEAKER_00 1:50

Correct. And the GitHub repo became the fastest in history to hit 50,000 stars two hours after publication.

SPEAKER_01 1:59

That's not just fast, that's a statement.

SPEAKER_00 2:02

Sigrid Jinn isn't some random developer. The Wall Street Journal featured him earlier this month in an article called The Trillion Dollar Race to Automate Our Entire Lives. And this is where the story gets wild. Quote, AI startup worker Sigrid Jinn single-handedly used 25 billion claw code tokens last year. At the time, usage limits were looser, allowing early enthusiasts to reach tens of billions of tokens at a very low cost.

SPEAKER_01 2:32

25 billion tokens. That's not casual use. That's living inside the product.

SPEAKER_00 2:40

Right. And it gets better. Quote, Jin flew to San Francisco in February for Claw Code's first birthday party, where attendees waited in line to compare notes with the founders. The crowd included a practicing cardiologist from Belgium who had built an app to help patients navigate care, and a California lawyer who made a tool for automating building permit approvals using Claw Code.

SPEAKER_01 3:05

So he was an insider, he knew the product intimately, he was invited to the birthday party.

SPEAKER_00 3:11

And then, when the source code leaked, he cloned it in two hours using a competitor's AI.

SPEAKER_01 3:18

That's not theft, that's competitive intelligence at AI speed.

SPEAKER_00 3:23

Here's where it gets legally fascinating. Clean room engineering used to be the safe way to clone a competitor.

SPEAKER_01 3:30

What is that exactly?

SPEAKER_00 3:32

You study their product, document how it works, then build your own version independently. It's legal because you can prove you didn't just copy their code.

SPEAKER_01 3:41

And that was legal because it took time. Months, years.

SPEAKER_00 3:46

You could prove the work was original? Right. But when AI can do the whole process in two hours, good luck proving that in court. The knowledge transfer happens so fast, the line between studied it and copied it completely disappears. So the legal framework just doesn't apply anymore.

SPEAKER_01 4:04

Correct. It was built for human speed. AI speed broke it. And here's the thing: this wasn't some obscure product. This was Anthropic, one of the biggest AI companies in the world. If they can't protect their flagship product, who can?

SPEAKER_00 4:20

Nobody. That's the point.

SPEAKER_01 4:22

Let's talk about the meta layer here. He used OpenAI's AI to clone Anthropic's AI.

SPEAKER_00 4:29

That's the whole competitive landscape right there. One AI company's product reverse-engineered another AI company's product in two hours.

SPEAKER_01 4:38

So if the AI companies are cloning each other, what chance does anyone else have?

SPEAKER_00 4:44

None. Unless you're the one doing the cloning.

SPEAKER_01 4:47

And that's the shift. The arms race isn't between you and your competitors anymore. It's between which AI tool they're using to clone you.

SPEAKER_00 4:56

Right. And the speed advantage is overwhelming. Traditional clean room engineering? Months. AI-powered clean room? Hours.

SPEAKER_01 5:06

So the old moats, patents, trade secrets, proprietary code, they're all gone.

SPEAKER_00 5:12

Not gone. Just irrelevant. Speed is the only moat now. If you can't move faster than someone can clone you, you're done.

SPEAKER_01 5:20

Alright, so what does this mean for a small business owner listening to this?

SPEAKER_00 5:24

Simple. If your software can be described, it can be cloned overnight. Meaning what exactly? Let's say you're paying$10,000 a year for a SAS tool. If you can describe what it does, really break down the features, the workflow, the data model, you might be able to build a custom version for your business in a weekend. Using the same AI tools this guy used. Exactly. Open AI Codex. Anthropic Claude. Any of the major coding assistants. You feed them the requirements, and they build it. So the flip side is this same power works for you too. Correct. You're not just vulnerable to being cloned. You can clone the tools you're paying for. That's the weapon, not a shield thing. Exactly. It's a weapon, not a shield. Use it accordingly.

SPEAKER_01 6:14

But let's be real for a second. Most small business owners aren't going to clone their own SaaS tools.

SPEAKER_00 6:21

True. But they need to understand the landscape, because their competitors might.

SPEAKER_01 6:27

And if your business model is built on proprietary software, you need to know that proprietary doesn't mean what it used to.

SPEAKER_00 6:36

Right. The question isn't can someone clone this? The question is, how fast can someone clone this?

SPEAKER_01 6:44

And the answer is getting faster every month. Weeks?

SPEAKER_00 6:49

Sometimes days. So what's the defense? Speed. You have to move faster than they can copy you. Constant iteration, constant improvement. The moment you stop, someone will catch up.

SPEAKER_01 7:03

Or you build around something that can't be cloned. Relationships, trust, brand, the stuff that takes time. Right.

SPEAKER_00 7:11

Because AI can clone features. It can't clone reputation.

SPEAKER_01 7:15

Alright, that's the story. One developer, two hours, and a legal framework that wasn't built for AI speed.

SPEAKER_00 7:22

The question isn't whether this keeps happening, it will. The question is, are you building or getting built around?

SPEAKER_01 7:31

I'm Michael. That was Frank. If this made you think, hit follow. We drop these every day.

SPEAKER_00 7:36

See you tomorrow.