MCP startup Runlayer accuses Rippling of stealing its product idea

We have all heard the warnings. You get the meeting with the enterprise giant, you show them your roadmap under an NDA, and six months later they announce a feature that looks exactly like your product. It is the classic startup nightmare. But in the world of Model Context Protocol (MCP), where the stakes for AI interoperability are skyrocketing, the recent legal clash between Runlayer and Rippling feels like a watershed moment for the ecosystem.

The Core of the Dispute

Runlayer, a startup focused on building a gateway and security layer for MCP, is suing Rippling, alleging that the HR and payroll giant effectively strip-mined their product ideas. According to the claims, Rippling engaged with Runlayer under the guise of a potential partnership or customer relationship, poked around the tech, and then pivoted to building a near-identical internal solution. This is not just a story about intellectual property; it is a lesson in the brutal reality of the AI infrastructure race.

For those not living in the developer docs, MCP is essentially the new standard for how AI models talk to data sources and tools. It is the plumbing of the agentic future. Runlayer thought they had found a niche by securing and managing these connections. Rippling, which manages a massive amount of sensitive employee data, clearly saw the value in that same plumbing. The question now is whether Rippling crossed a legal line or if this is just the inevitable gravity of 'buy vs. build' playing out in the most aggressive way possible.

The Founder Perspective: The MNDA Illusion

As a founder, you are often told that a Mutual Non-Disclosure Agreement (MNDA) is your shield. The reality is that for a startup with limited capital, an MNDA is often just a ticket to a courtroom you cannot afford to stay in. In this case, Runlayer is taking the fight to a billion-dollar incumbent. This is a high-stakes gamble that signals how vital they believe their specific architecture is to the future of the AI stack.

We have seen this movie before in the early days of SaaS and cloud. A small player identifies a friction point, builds a clever solution, and then the dominant platform provider realizes they cannot leave that piece of value to a third party. The difference here is the speed. In the AI era, the transition from 'discovery call' to 'competitor' is happening in months, not years.

Why MCP is the New Battleground

Why fight over an MCP gateway? Because he who controls the context controls the agent. If you are a company like Rippling, your entire moat is built on the data you hold—payroll, benefits, identities. As AI agents become the primary way users interact with that data, Rippling needs to ensure that the gateway those agents use is secure, compliant, and internal. From their perspective, letting a startup like Runlayer sit in the middle of their data flows might have looked like a strategic liability.

However, if the allegations hold water, the method of execution matters. There is a right way to decide to build internally, and there is a way that involves using a startup as a free R&D department. For builders in the crypto and AI space, this is a loud reminder that your most valuable asset is not just your code, but the specific 'know-how' of how you solved the edge cases—the stuff you usually talk about in those early 'vision' meetings.

The hard truth for AI founders: If your product can be replicated by a dedicated internal team at a major platform after three demo calls, you don't have a moat. You have a feature.

What This Means for the Builders

If you are building in the AI infrastructure space right now, you need to change your posture. The era of 'open kimono' meetings with enterprise giants is over. Here is how I look at the landscape post-Runlayer:

• Functional Siloing: Don't show the whole machine. Show the output. If you are pitching a gateway, talk about the results, not the specific routing logic or the way you handle state across MCP servers.
• Speed is the Only Defense: If a giant is going to steal your idea, you have to be two versions ahead by the time they ship. If your roadmap is stagnant, you are a sitting duck.
• The Enterprise Ego: Never underestimate the 'not invented here' syndrome. Major tech firms are currently terrified of losing their data dominance to AI startups. They are more likely to clone you than buy you in this current climate.

The Skeptical Take

While it is easy to paint Rippling as the villain here, there is a counter-argument. Large companies have been building internal gateways and middleware for decades. MCP is an open standard. Is it possible that Rippling was already thinking along these lines and the talks with Runlayer simply confirmed their internal thesis? Distinguishing between 'stolen IP' and 'market convergence' is what the lawyers will spend millions on.

Regardless of the legal outcome, the reputational damage to enterprise firms who 'cradle-snatch' ideas is real. If the developer community starts to see a specific big tech player as a predator rather than a partner, they will stop building on their APIs. In the long run, burning the bridge to the startup ecosystem is a net loss for the incumbent.

The Final Word

The Runlayer vs. Rippling case is a wake-up call for the MCP ecosystem. We are moving away from the 'everyone is just experimenting' phase into a period of consolidation and defensive maneuvering. If you are a founder, keep your head down and your secrets closer. Success in AI isn't just about having the best model or the cleanest gateway; it's about surviving the interest of the giants who want what you've built.

The Takeaway: An MNDA is not a brick wall; it is a paper fence. Builders should treat every enterprise partnership discussion as a potential competitive intelligence gathering mission until the wire transfer clears.

Read the original at TechCrunch AI →