Runlayer and Rippling have ended a brief legal fight over software that controls how enterprise AI agents connect to company data. Court documents reviewed by TechCrunch show that both sides dropped their lawsuits on Wednesday night, with no settlement, payment or legal fees changing hands.
The dispute lasted only a few weeks, but it captured a wider risk facing young infrastructure companies. A large potential customer can spend months evaluating a product, learn how the category works and then decide to build a competing system instead. That possibility is especially acute in AI software, where products and markets are developing quickly.
A long evaluation ended with a competing launch
Runlayer launched in November 2025 and has since raised $42 million. Its product includes a gateway for the Model Context Protocol, or MCP, a standard used to connect AI agents with external tools and data. Such a gateway can route requests, apply role-based access controls and provide logs and observability for administrators.
According to Runlayer’s lawsuit, Rippling tested its MCP gateway for more than a year but ultimately did not sign a commercial agreement. Rippling then developed its own gateway. The complaint cited a Rippling employee who allegedly described that product as a clone, although the source report does not establish that statement as an admission of unlawful copying.
Runlayer sued over alleged contractual violations. Rippling responded with patent claims of its own. The competing filings transformed a prospective customer relationship into an unusually public conflict between an established business-software company and a much younger startup.
After three weeks of discovery, Runlayer voluntarily dismissed its case and Rippling dismissed its counterclaims. The court filings say neither company paid the other. That outcome closes the litigation without resolving the competing accounts of how the product evaluation unfolded.
The underlying gateway market keeps moving
Rippling released the MCP gateway at the centre of the dispute immediately after the lawsuits were dropped. The company has been expanding into AI infrastructure, including model routing, token-spend management and agent security. Runlayer, meanwhile, packages its gateway with controls intended to identify and manage unsanctioned or “shadow” agents.
The overlap helps explain why the disagreement mattered. An MCP gateway is not simply a connector. In an enterprise environment, it can become a control point for deciding which agents may reach sensitive systems, what actions they may take and how those actions are recorded. A vendor that owns that layer may gain a valuable position in a company’s AI stack.
Neither the launch nor the abandoned lawsuits prove that one product is equivalent to the other. TechCrunch’s report describes the allegations and the sequence of events, but it does not offer a technical comparison or a judicial finding on ownership. The patent counterclaims were also withdrawn before a court tested them.
What is verified is narrower: Rippling evaluated Runlayer’s technology, the companies later offered competing gateways, both sued, and both walked away after a short discovery period. The absence of a settlement means the commercial and technical questions remain unanswered rather than decided.
Enterprise sales can create a strategic exposure
For founders, the episode illustrates a tension built into long enterprise sales cycles. A startup often needs to demonstrate its product deeply enough for security, procurement and engineering teams to assess it. That process can expose the buyer to the product’s architecture, capabilities and market opportunity before any contract is signed.
The risk is not unique to AI, but rapid development lowers the time required for a capable company to enter an emerging category. A prospective customer may also have distribution, integrations and engineering resources that a startup cannot match. Legal protections can define confidentiality and permitted use, yet enforcing them may consume time and capital without producing a clear result.
The Runlayer-Rippling dispute does not establish that extensive evaluations should be avoided. It does show why startups need to decide carefully what they reveal, when they reveal it and which contractual safeguards apply during a trial. In a fast-moving AI infrastructure market, a sales process can become competitive intelligence long before it becomes revenue.
