When Enterprise Trials Turn Into Courtroom Battles

The Allegation

Runlayer has filed suit against Rippling, claiming the HR software company built a competing Model Context Protocol gateway using intellectual property shared during an extended product evaluation. The complaint describes nearly a year of technical collaboration, during which Runlayer disclosed product roadmaps and source code under mutual non-disclosure and trial agreements that explicitly prohibited copying or creating derivative works.

According to the filing, the partnership fell apart over pricing disagreements. Shortly after Runlayer terminated the trial, founder and CEO Andrew Berman received a text from someone described as a "Rippling insider" warning that an internal project was underway to build "essentially a clone" of Runlayer's offering, characterized as "almost a 1 to 1 copy."

Rippling has confirmed it will launch its own MCP gateway but denies any misuse of Runlayer's intellectual property. A company spokesperson characterized the lawsuit as a "panicked effort to avoid competition" and stated that Rippling developed its product "using only our proprietary information."

The Structural Problem

At DailyTechWire, we've tracked enterprise AI infrastructure deals across the region for the past eighteen months, and this dispute illustrates a structural tension that has only intensified as the MCP ecosystem matures. Startups building foundational tools face customers with the engineering capacity to replicate those tools in-house. The more sophisticated the buyer, the greater the risk.

Enterprise sales cycles in complex infrastructure routinely extend six to twelve months. Vendors must demonstrate deep integration capability, which means sharing architectural decisions, API designs, and often substantial portions of source code. Trial agreements and NDAs offer legal protection on paper, but enforcing them requires expensive litigation with uncertain outcomes, and the damage to reputation and fundraising momentum can precede any verdict.

Runlayer has retained Sullivan & Cromwell, a signal that the startup is prepared for a protracted fight. The choice of counsel lends the case credibility in the same way a lead investor from Sequoia or a16z lends a funding round credibility. But legal firepower does not guarantee success, and discovery will ultimately determine whether Rippling's gateway genuinely derives from independent work or relies on material disclosed during the trial.

MCP Gateway Economics

Model Context Protocol gateways sit at a strategic choke point in the emerging AI stack. Anthropic open-sourced the MCP standard in November 2024, creating a common interface for AI models and agents to securely access external data sources and services. Gateway products layer on authentication, access control, observability, and governance features that enterprises require when deploying agents at scale.

Runlayer launched its commercial gateway in mid-2025 and has raised a total of forty-two million dollars from investors including Khosla Ventures and Felicis. The company entered a market that was still taking shape, with relatively few competitors and strong tailwinds from accelerating agent adoption.

That landscape has shifted. Multiple vendors now offer MCP gateways, and the core protocol's open-source nature lowers the barrier for any engineering team to build a custom implementation. For a company like Rippling, which already manages sensitive HR and payroll data across thousands of customers, the strategic case for owning the MCP layer is clear. The question is whether the technical path to that ownership crossed legal boundaries.

Build Versus Buy in the Age of Agents

The dynamics here extend beyond one lawsuit. AI infrastructure startups face a category of buyer that is simultaneously their best prospect and their greatest existential threat. Large technology companies with internal AI teams can evaluate a product deeply enough to understand its full value, but that same depth of evaluation equips them to build an alternative.

Traditional enterprise software mitigated this risk through network effects, switching costs, and proprietary data moats. MCP gateways have weaker versions of each. The protocol itself is standardized and open. Integration work is non-trivial but not insurmountable for a well-resourced engineering organization. Data flows through the gateway but does not accumulate within it in ways that create lasting lock-in.

This creates a compressed window for vendors to prove differentiated value before customers decide to build. Runlayer's complaint suggests it was operating within that window when the relationship with Rippling soured. Whether the startup can prove that Rippling's subsequent product derives from protected disclosures will shape how other infrastructure companies approach enterprise trials going forward.

What Comes Next

Litigation of this kind moves slowly. Discovery will likely center on Rippling's internal communications, development timelines, and architectural decisions. Runlayer will need to demonstrate that specific elements of Rippling's gateway correspond to proprietary designs or implementations shared during the trial, rather than to publicly available MCP documentation or independent engineering judgment.

Even if Runlayer prevails, the remedy may be limited. Injunctions are difficult to obtain in software disputes, and damages hinge on proving both misappropriation and quantifiable harm. Meanwhile, Rippling will continue building and shipping its gateway, and customers evaluating MCP solutions will weigh the legal uncertainty alongside technical and commercial factors.

For founders in the AI infrastructure space, the case serves as a sobering reminder that deep collaboration with sophisticated buyers carries risk that contracts alone cannot fully mitigate. The most effective defense may be speed: shipping differentiated features faster than customers can replicate them, building integrations that require sustained investment to maintain, and cultivating relationships that make in-house development politically and economically unattractive.

The enterprise sales playbook has always required balancing transparency with protection. In a market where the underlying protocols are open and the buyers are engineers, that balance has never been harder to strike.