Runlayer, a startup selling a secure Model Context Protocol (MCP) gateway, has filed a lawsuit against HR software company Rippling, accusing it of stealing its product idea. According to TechCrunch AI, which reviewed the complaint, Runlayer claims Rippling ran a lengthy product trial as a prospective customer, saw everything under the hood, then built a near-identical clone of its own.
What stands out here isn’t just the courtroom drama. It’s the window this case opens into how risky it has become to sell AI infrastructure to enterprise buyers who can build the same thing themselves.
What happened
Here’s the sequence as laid out in the complaint, per TechCrunch AI:
- Rippling evaluated Runlayer as a potential customer. Runlayer says the trial spanned “nearly a year of intensive engineering collaboration.”
- During that trial, Runlayer shared its product roadmap and its actual source code.
- Both sides signed a mutual NDA. Rippling also signed a trial agreement barring it from copying Runlayer’s IP or making derivative works, standard boilerplate in enterprise software deals.
- The two couldn’t agree on price, so Runlayer ended the trial.
- Shortly after, Runlayer alleges a “Rippling insider” texted CEO Andrew Berman to warn him of an internal project to build “essentially a clone” that was “almost a 1 to 1 copy of Runlayer.”
Runlayer’s suit claims trade secret misappropriation, unfair competition, and breach of contract.
Rippling pushes back hard
Rippling confirmed to TechCrunch AI that it is launching its own MCP gateway, but flatly denies misusing anyone’s intellectual property.
“Runlayer’s panicked effort to avoid competition by fabricating claims is not an effective way to deal with its business failures,” a spokesperson said. “Rippling is launching a superior product for connecting AI tools to business data using only our proprietary information. We have every reason to win in this market.”
Runlayer isn’t going in light. It’s retained Sullivan & Cromwell, a white-shoe firm. That doesn’t decide the case, but as TechCrunch AI notes, a marquee law firm lends a lawsuit some credibility, at least optically, the same way a big-name VC lends a startup credibility.
Why this matters
This is significant because it exposes a structural trap in selling complex AI infrastructure into the enterprise. Enterprise sales take a long time to close, often because they hinge on deep, hands-on trials. Those trials require you to show the customer how your product actually works. When the customer is another tech company with strong engineers, you’ve just handed a potential competitor a blueprint.
A quick primer on the technology at the center of this. MCP, or Model Context Protocol, is a standard that lets AI models and agents securely pull in outside data and tools. Anthropic launched it as an open source protocol in November 2024, and it’s now one of the basic building blocks of AI interoperability. MCP gateway products sit on top, adding control, security, and management features, especially for governing agents.
The catch: because MCP itself is open source and the gateway category is young, the field is crowding fast. Runlayer launched its product in the middle of last year and has raised $42 million total, including from Khosla Ventures and Felicis. That funding didn’t buy it a moat.
What to watch next
For founders selling AI infrastructure, the practical takeaways are blunt:
- NDAs and anti-copying clauses are standard, but this case tests whether they actually hold when a well-resourced buyer decides to build in-house.
- Guard what you expose in a trial. Source code and roadmaps are the crown jewels, and once shared, they’re hard to un-share.
- Expect more of these disputes as AI building blocks commoditize and big customers weigh buy versus build.
Both sides are stuck. Runlayer has to prove Rippling’s product came from its secrets, not independent work, which is a high bar. Rippling has to show it built a genuinely separate product after a year inside Runlayer’s code. As TechCrunch AI puts it, both are caught between a rock and a hard place.
Expect this one to shape how AI infrastructure startups run their enterprise trials going forward. Full details are available at the original TechCrunch AI report.