Vibe coding has become a serious topic in the startup world — and the most recent case involves Corgi, a Y Combinator-backed insurtech that finds itself at the center of a major controversy.
The drama kicked off when Papermark, the creator of an open source data room software, publicly accused Corgi of copying their product and relaunching it as their own. The accusation came straight from Papermark co-founder Marc Seitz, via X, complete with screenshots — showing nearly identical text and features across both products.
Corgi flatly denies any code theft. In a statement to the press, the company said no Papermark code was used. But the story doesn’t end there 👀
What this episode brought to light goes far beyond a corporate feud. It raises a question that’s going to make the industry increasingly uncomfortable: if an AI can replicate the look, the language, and every feature of a product without copying a single line of code, what does it actually mean to copy something?
That’s the knot Corgi helped tighten — and the one we’re going to untangle right here. 🧵
What actually happened between Corgi and Papermark
Papermark is a well-known open source project in the tech community. It works as a data room software — basically, a tool for securely sharing documents. This type of tool is famously used by startups to pitch venture capital investors and share supporting materials during due diligence processes. Because it’s open source, its code is publicly available — and that’s exactly the centerpiece of this whole story.
Marc Seitz, Papermark’s co-founder, posted side-by-side comparisons on X between the Papermark product and the data room Corgi just launched, branded as Dataroom. The screenshots showed nearly identical text for the same features, word for word. Seitz went further, calling Corgi’s new product a copyright and license infringement, labeling the situation as fraud.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The community reaction was immediate. Developers, founders, and tech enthusiasts started debating the case with an intensity you rarely see in corporate disputes in this space. Hundreds of comments and countless indirect posts flooded X.
Corgi co-founder and CEO Nico Laqua saw the tweet and promised to investigate. Shortly after, he responded publicly with a full denial, showing that the code was different between the two products. While he firmly pushed back on the license violation accusations — arguing that copying my style is a different claim than stealing corporate code — he admitted that relying on vibe-coded design led to the replicated features.
Looking back, we should have leaned more into our own language and visual choices instead of pulling references from existing products in the market, and that’s on us, Laqua wrote. A Corgi spokesperson confirmed that the problematic features were built through vibe coding and said they had already been changed, downplaying the situation. According to the spokesperson, the issues were limited to visual elements on two peripheral settings pages, which were updated immediately.
Vibe coding: the new way to build software that nobody knows how to regulate
To understand why this case is so complicated, you need to understand what vibe coding actually looks like in practice. The idea is straightforward: instead of writing code by hand, the developer describes what they want in plain language — sometimes in pretty vague terms — and the AI generates the code. The result can be surprisingly functional, even when the person writing the prompts doesn’t have deep technical expertise.
The problem starts when the AI is trained on massive datasets that include public repositories — including, of course, open source projects like Papermark. When you ask an AI to build a modern data room with specific features, it’s going to pull references from what it has already learned. And if the best available example for that type of product happens to be Papermark’s code, there’s a good chance the generated output will look a whole lot like it — without the user having manually copied a single line.
That’s the point that makes this debate so thorny. From a technical and legal standpoint, Corgi might very well be telling the truth when they say they didn’t copy code. But from an ethical standpoint — and especially from the perspective of someone who spent years building an open source project — the feeling of having your work repackaged without any credit is exactly the same. Vibe coding creates a kind of gray zone where intellectual property loses its clear boundaries, and nobody quite knows how to deal with it yet.
As Dan Barrett, also a Y Combinator alum and founder of the agent operating system OpenProse, pointed out: in a world where a bot can trivially copy, one-to-one, the structure of something even if the character-level code is different, what makes one thing unacceptable and the other not? Existing IP law, inherited from an old world? Isn’t there some bigger principle at play here?
The open source dilemma in a generative AI world
The open source movement was built on a foundation of collaboration and reciprocity. When a project publishes its code openly, there’s an implicit expectation — and often an explicit one, through licenses — that anyone benefiting from that work will, at the very least, credit those who came before. Licenses like MIT, Apache, and GPL have different nuances, but they all start from the idea that using code requires following certain minimum rules around attribution and sharing.
What generative AI does is completely scramble that playbook. When a language model learns from thousands of open repositories and then generates new code that resembles those repositories, who’s the author of the result? The AI? The user who wrote the prompt? The people who originally wrote the code that served as training data?
It’s worth noting that this case is different from the controversy involving PearAI, another Y Combinator startup that in 2024 admitted to cloning another open source project and relaunching it under its own license. In that episode, there was a clear admission of cloning. In Corgi’s case, the company maintains that not a single line of code was copied — which makes everything morally ambiguous and, according to experts, increasingly common.
Corgi’s litigious reputation enters the picture
Corgi is now scrambling to clean up any damage to its image. The company confirmed it sent a cease-and-desist letter to Seitz demanding the removal of the original tweet. The founder of Hello World Cafe, which partially competes with Corgi’s coffee shop business, also said he received a legal notice from the company’s lawyers over a tweet joking about the Dataroom controversy.
This isn’t the first time Corgi has been accused of aggressive legal tactics. In May, competitor Matcha accused the company of bullying behavior in a dispute that unfolded alongside a separate lawsuit. The startup, only two years old, has already sued several former employees and built a growing reputation for being litigious.
Laqua also went viral recently for comments on investor Harry Stebbings’ podcast, where he said he expects his employees to work seven days a week. According to him, what you can do in five days, I guarantee, you’ll do more in six and seven. This, of course, is the classic hustle-culture fallacy of startups. Decades of research consistently conclude that human productivity isn’t an equation that scales linearly with hours worked. While short sprints can be effective for specific problems, studies show that as a routine, more work hours actually reduce productivity instead of boosting it.
Massive fundraising and breakneck growth
The startup has also made headlines for the speed at which it has been raising money, with increasingly higher valuations — even by AI industry standards. Last month, Corgi raised a $106 million Series B1 round, valuing the company at $2.6 billion — just three weeks after announcing a $160 million Series B at a $1.3 billion valuation, and four months after its $108 million Series A. The company also operates a 24-hour coffee shop, with plans to open more locations, according to Laqua.
What this episode changes for startups going forward
Regardless of how this specific dispute plays out, it has already left a mark on the conversation about how startups should use vibe coding tools in product development. The ease that AI offers for generating functional interfaces, replicating UX patterns, and cloning complex workflows in a matter of hours is tempting — especially for small teams with limited resources and tons of pressure to ship fast.
But the Corgi episode shows that this speed comes with a reputational cost that can be pretty steep. Even if you didn’t copy anything intentionally, if the end result of your product looks too much like another existing product — especially an open source project with an active community — you’re going to face tough public questions that are hard to answer. And in the age of social media, an accusation with side-by-side screenshots spreads way faster than any technical explanation.
What the market is starting to realize is that vibe coding doesn’t eliminate the responsibility of the people building the product. It changes the process, but it doesn’t change the need to understand where the references the AI is using come from, what licenses are involved, and how to make sure the final result respects the work of those who laid the foundation. That requires a level of maturity still being developed in the startup ecosystem — and cases like this one are a painful but necessary part of that growing process. 🐾
The debate over the ethical and legal boundaries of vibe coding is just getting started, and Corgi may have gone down in history as the first major example of how this territory still doesn’t have clear enough rules to protect anyone — not the accusers, and not the accused.
