Runlayer Sues Rippling Over Alleged MCP Gateway Clone After Year-Long Trial

Runlayer, a startup selling a secure Model Context Protocol gateway for AI tools and agents, has sued HR software company Rippling, alleging that Rippling copied its product after an extended enterprise trial in 2026. The case matters because it highlights how quickly AI infrastructure competition is intensifying — and how risky long, hands-on evaluations can become for startups selling into large companies.

According to the complaint described by TechCrunch, Runlayer says Rippling spent nearly a year evaluating the product, gaining access to detailed roadmap information and source code under NDA and a product trial agreement, before walking away and later preparing to launch a similar offering. Rippling denies the accusations and says it is building its own gateway using proprietary work.

What Runlayer says happened

Runlayer’s core allegation is straightforward: it believes Rippling used the course of a long sales-and-testing process to learn enough about the startup’s technology to recreate it internally. The startup says the relationship began as a normal enterprise product trial, but evolved into a deep engineering collaboration that went far beyond a typical demo or pilot.

In the complaint, Runlayer argues that Rippling had access to sensitive materials while evaluating the product, including portions of the company’s roadmap and source code. The startup says both companies signed a mutual nondisclosure agreement, and that Rippling also agreed to a trial contract prohibiting copying, derivative works and other forms of intellectual-property misuse.

The lawsuit says the trial stretched over “nearly a year,” a length of time that reflects the realities of enterprise software sales, where buyers often want extensive security reviews, custom integration work and proof that a new tool can operate inside complex corporate systems.

Why the lawsuit matters for AI infrastructure

The dispute goes beyond two startups. It speaks to a broader tension in the AI market: the more specialized the infrastructure product, the easier it may be for a technically sophisticated prospective customer to decide it can build something similar itself.

That risk is especially high in the market for Model Context Protocol, or MCP, gateways. These products sit between AI models and the outside systems they need to access, such as business databases, internal tools and cloud services. They are designed to make those connections safer, more governed and easier to manage as companies begin deploying agents that can take actions on behalf of users.

For startups, the upside is clear: if a customer adopts the product, the company can become deeply embedded in enterprise workflows. The downside is equally clear: a large customer with enough engineers may decide the idea is valuable enough to replicate on its own, especially if the product concept is relatively new and the market is still defining standards.

How the trial allegedly turned into a copycat dispute

Runlayer says the relationship with Rippling ended when the two sides could not agree on pricing. After that, the startup alleges, an employee with knowledge of Rippling’s internal plans contacted Runlayer founder and chief executive Andrew Berman with a warning that the company was building something nearly identical.

Runlayer claims the message suggested an internal project that would become “almost a 1 to 1 copy” of its own gateway. Based on that, the startup says Rippling’s forthcoming product must have been built using Runlayer’s confidential information and trade secrets rather than developed independently.

The legal claims include trade secret misappropriation, unfair competition and breach of contract. Those are common causes of action in disputes over software ideas, but they can be difficult to prove unless the plaintiff can show both access to protected information and a strong resemblance between the products or source code.

Runlayer’s position is that an extended enterprise evaluation gave Rippling access to its roadmap, engineering details and code, then the prospective customer re-created the product after the commercial talks fell apart.

What Rippling says in response

Rippling has confirmed it is working on its own MCP gateway, but the company rejects the accusation that it stole Runlayer’s intellectual property. In a statement to TechCrunch, a Rippling spokesperson called the lawsuit a manufactured attempt to block competition and said the company’s product is being built from its own internal work.

Rippling’s public response frames the case as a standard business dispute rather than a misappropriation claim. The company says it has every reason to win in the market and describes its upcoming tool as a better way to connect AI systems to business data.

That reaction is important because it suggests the companies are likely to fight not just over facts, but over a larger narrative: whether this is a case of copying, or simply what happens when a customer learns enough from an evaluation to pursue a similar product on its own.

Rippling’s position, as described to TechCrunch, is that it is launching a superior MCP gateway based on proprietary information and that Runlayer’s lawsuit is an attempt to avoid competition.

What is an MCP gateway?

An MCP gateway is a control layer for the Model Context Protocol, which Anthropic introduced as an open-source standard in November 2024. MCP allows AI models and agents to connect securely to outside data sources and software tools without needing custom integrations for every system.

In practice, a gateway sits in front of those connections and helps enterprises manage permissions, auditing, security policies and access controls. That makes it especially relevant for businesses deploying AI agents that may need to retrieve documents, query internal systems or complete tasks across multiple applications.

Because MCP is becoming a foundational layer for AI interoperability, companies building products around it are competing in a fast-moving and increasingly crowded category. Early movers like Runlayer are not just selling software; they are trying to define the control plane for a new generation of AI workflows.

Why enterprises care

Enterprises care because agentic AI is only useful if it can connect safely to internal data and tools. They also care because every new integration creates security and governance concerns, especially when systems can take action rather than simply answer questions.

An MCP gateway can help address those concerns by centralizing controls and creating a more visible path for data access. That is one reason vendors in this space are pitching themselves as essential infrastructure rather than optional add-ons.

Who is Runlayer, and how well funded is it?

Runlayer is a relatively young startup that launched its product in the middle of last year and has raised a total of $42 million, according to the source material. Its backers include Khosla Ventures and Felicis, two well-known venture firms with a track record of investing in technical infrastructure companies.

That funding does not guarantee success in court, but it does show the company has enough momentum to attract serious capital in a market where investors are betting on tools that make AI deployment safer and more manageable.

The fact that Runlayer retained Sullivan & Cromwell, a prominent law firm, also underscores how seriously it is treating the case. A high-profile legal team does not prove the merits of a claim, but it can indicate that the plaintiff expects a fight and wants to signal confidence to customers, investors and rivals.

Why enterprise sales can become a trap for startups

Long enterprise sales cycles are a double-edged sword. On one hand, they can lead to large, sticky contracts that justify the time spent on demos, pilots and technical support. On the other hand, they can give buyers unusually deep visibility into how a product works.

That dynamic becomes even more fraught when the buyer is itself a technology company with engineering resources. In those cases, the line between evaluation and reverse engineering can become a disputed gray area, especially when the product sits in a still-forming category where technical differentiation is hard to prove from the outside.

Runlayer’s lawsuit serves as a warning that startups selling AI infrastructure to sophisticated enterprises may need to think carefully about what they disclose during pilots, how they limit access and how they structure agreements around trial use.

Common risks in deep product trials

• Extensive access to technical documentation and architecture details
• Exposure of roadmap items before a commercial contract is signed
• Source code reviews that may reveal implementation choices
• Large customers deciding to build internally after learning the workflow
• Difficulty proving where inspiration ends and copying begins

How crowded is the MCP gateway market?

The market is becoming crowded quickly. MCP itself was only introduced in late 2024, but it has already become a key layer in the broader AI stack, and products built around the protocol are multiplying.

Runlayer’s case shows how quickly category leaders can face pressure from larger companies with adjacent needs. Once a standard becomes important enough for enterprise adoption, large vendors often decide they do not want to depend on a startup for a core integration layer — especially if the startup is early and the market is still unsettled.

That creates a race in which startups must balance speed, secrecy and customer trust. If they are too closed, enterprises may hesitate to buy. If they are too open, they may hand potential rivals the blueprint for their business.

What happens next?

The next stage will likely be a legal contest over evidence, not just headlines. Runlayer will need to show that Rippling had access to protected information and that the internal project crossed the line from independent development into unlawful copying.

Rippling, meanwhile, is likely to argue that MCP gateways are a natural product category for a company with its own AI ambitions and that any similarities reflect the demands of the market rather than theft. If the case moves forward, discovery could reveal how much the companies shared, what was written in the agreements and whether any internal messaging supports Runlayer’s version of events.

Even without a court ruling, the dispute may shape how other startups approach enterprise trials in the AI infrastructure market. Founders selling into large, well-resourced buyers may become more cautious about what they reveal, how they license trials and how much technical detail they share before signing a real contract.

For now, the lawsuit stands as a reminder that in fast-moving AI markets, the value of an idea may depend as much on execution and trust as on novelty. And when a customer can also become a competitor, the risks of selling the blueprint can be as significant as the reward of winning the deal.

Timeline of the dispute

1. Mid-2025: Runlayer launches its MCP gateway product and begins building traction in enterprise AI infrastructure.
2. Over roughly a year: Rippling evaluates Runlayer in a lengthy enterprise trial, according to the complaint.
3. During the trial: Runlayer says it shared roadmap information and source code under NDA and trial restrictions.
4. After pricing talks fail: Runlayer ends the trial.
5. Shortly afterward: Runlayer says a Rippling insider warns the startup about an internal clone project.
6. July 28, 2026: TechCrunch reports that Runlayer has filed suit against Rippling.

Bottom line

This is more than a one-off legal fight. The case captures the pressure points in enterprise AI sales, where startups are trying to commercialize new infrastructure while large customers are increasingly capable of building similar tools themselves. Whether Runlayer wins or not, the lawsuit is likely to become a reference point for the next wave of AI infrastructure deals.