Runlayer and Rippling Drop Their AI Gateway Lawsuits, Exposing a Faster, Riskier Startup Market

Runlayer and Rippling have both dropped their lawsuits against each other, ending a brief but noisy legal fight over AI gateway technology with no settlement, no damages and no attorneys’ fees paid, according to court filings reviewed by TechCrunch. Rippling then immediately released its own MCP gateway product, turning the dispute into a vivid warning for founders building in the fast-moving AI infrastructure market.

The episode matters because it shows how quickly a startup’s assumed customer can become a competitor in the age of generative AI. It also highlights how enterprise software buying cycles, which can stretch for months, may no longer match the pace at which AI products can be built, copied and launched.

What happened between Runlayer and Rippling?

Runlayer accused Rippling of using extended product testing to learn from its technology, only to walk away and ship a competing service. Rippling answered with its own patent claims. By Wednesday night, both companies had withdrawn their cases, ending the courtroom fight without a financial deal.

The legal dust-up lasted only weeks in public view, but it generated outsized attention because it captured a larger concern across the AI startup ecosystem: when the cost of building software falls and product cycles compress, competitive boundaries become less predictable.

How the dispute started

According to Runlayer’s complaint, the startup spent more than a year letting Rippling test its MCP gateway while the two engineering teams worked closely together. But Runlayer says Rippling never became a paying customer. Instead, one of Rippling’s employees allegedly texted founder Andrew Berman to say the company was building its own gateway product and intended to launch it. That message, Runlayer said, described the Rippling version as a clone.

Runlayer then argued that Rippling had broken contractual promises tied to the product evaluations. Rippling responded by filing its own suit, alleging patent infringement and suggesting Runlayer’s claims were overstated. Runlayer viewed that counterattack as an effort to drive up legal costs and pressure it to back down.

Runlayer’s account framed the encounter as a cautionary example of what can happen when a startup gives a prospective enterprise buyer unusually deep access to a young product before a commercial relationship is finalized.

Why does this matter for AI founders?

Because the AI market is forcing a rethink of how startups structure enterprise trials. In traditional software sales, long technical evaluations could be worth the wait because product categories changed slowly and moats were clearer. In AI, a six-month pilot can be enough time for a buyer to decide to build a competing feature in-house.

That shift creates new risks for founders who rely on hands-on demonstrations, sandbox access, and close collaboration with large enterprises. The more deeply a startup embeds itself in a customer’s workflow during evaluation, the more it may be exposed to imitation, switching, or even direct competition.

The Runlayer-Rippling clash is not just about one gateway product. It is about whether early-stage vendors can still use the same sales motion they did in the previous software cycle, or whether they need much tighter limits, faster contracting and stronger technical safeguards.

What is an MCP gateway?

An MCP gateway is a control layer that sits between AI agents and an enterprise’s internal systems. Its job is to handle requests securely when an AI tool needs data from software such as recruiting platforms, payroll systems, customer databases or internal dashboards.

Instead of allowing an agent direct access to company systems, the gateway mediates the request, checks permissions and returns only the information the user or agent is allowed to see. That design can also support logging, audit trails, and role-based access control so managers, interns and other staff members receive different levels of access.

In practical terms, if a recruiter asks an AI agent for the top candidates in an open role, the gateway can fetch those records from the recruiting system without exposing broader database access. The point is not just convenience; it is also security, governance and visibility.

Key functions of an AI gateway

• Securely routes AI agent requests to enterprise data sources
• Applies role-based permissions before data is returned
• Tracks usage for logging and compliance
• Supports observability and audit trails
• Can limit what models and tools employees use

Who are Runlayer and Rippling?

Runlayer is a young startup that came out of stealth in November 2025 and has raised $42 million from investors including Khosla Ventures’ Keith Rabois and Felicis. It is led by Andrew Berman, a three-time founder whose previous companies include baby-monitor maker Nanit and the AI video-conferencing startup Vowel, which was acquired by Zapier in 2024.

Rippling, by contrast, is best known for payroll and benefits software. Over time, it has expanded into a broad workplace operations platform, and now it is pushing further into AI infrastructure and security tools. Its latest gateway product gives it a foothold in the same space Runlayer is targeting, along with rivals such as Docker and Amazon Bedrock.

The overlap is striking because it shows a large incumbent moving rapidly into a category that a startup was still trying to define. That is exactly the kind of situation that can turn a customer relationship into a competitive one almost overnight.

What does Rippling’s new product do?

Rippling’s gateway is designed to route AI requests across different models while also showing token spend by employee in a dashboard. That makes it useful not only for control and compliance, but also for cost management, especially as enterprises try to understand who is using AI and how much it is costing them.

By entering this market, Rippling is competing with vendors focused on routing, observability and AI security. The company’s pitch now extends beyond workforce management into the emerging infrastructure layer around agents and enterprise AI use.

Rippling’s launch suggests that large platform companies no longer want to sit on the sidelines while startups define the rules of AI access and governance.

How did the lawsuits end?

They ended quietly in legal terms, even though the rhetoric around them was loud. Runlayer dismissed its claims after three weeks of discovery. Rippling then dropped its counterclaims as well. No settlement was announced, and court records indicate neither side paid the other’s legal fees.

That outcome is unusual given how aggressively the dispute was framed. In many commercial cases, at least one side walks away with a payment, a licensing agreement or some formal restriction. Here, both companies appeared to prefer disengagement over a negotiated resolution.

Timeline of the dispute

Why is the case a warning about enterprise sales in AI?

It is a warning because the enterprise buying process can become obsolete faster than many founders expect. A startup may enter a pilot assuming that a long test will end in a contract, but the customer’s internal roadmap can shift, especially if the product is easy to understand, replicate or absorb into a bigger platform.

AI amplifies that problem. Model orchestration, access control and agent routing are important, but the underlying product concepts are often easier to explain and copy than older enterprise systems. That makes technical diligence both essential and dangerous for startups that have not yet established strong commercial protections.

The takeaway is not that founders should avoid enterprise pilots. Rather, they should design them more carefully, with tighter scopes, clearer usage boundaries and legal terms that reflect the speed of AI competition.

What broader market trends does this reflect?

The fight sits at the intersection of three trends: the rush to build AI infrastructure, the compression of product cycles and the scramble by incumbents to reposition themselves around AI. Each of those trends increases competitive pressure.

Runlayer is trying to sell security and governance for agents, a category that barely existed in the conventional enterprise stack. Rippling, meanwhile, is using its distribution and workplace software footprint to move into AI operations tools. Both are betting that whoever controls the gateway layer will shape how companies govern the next generation of software agents.

That is why the dispute drew attention well beyond the two companies. It illustrates how a new product layer can become strategically important almost as soon as it appears, and how swiftly a major platform can enter the same market.

Where Rippling competes now

• AI gateway routing and model selection
• Employee-level token spending visibility
• Enterprise AI security and access control
• Agent governance and observability

What founders should learn from the dispute

Founders should assume that every long pilot could expose product strategy, engineering details and market positioning to a party that might later decide to build the same thing. That does not mean every enterprise prospect is a threat, but it does mean trust should be paired with technical and legal discipline.

Companies building AI infrastructure may need to rethink how much access they grant before signing a deal, how long they allow paid or unpaid evaluations to run and what contractual language protects them if a prospect becomes a rival.

The other lesson is strategic: startups should expect incumbents to move faster than before. In a market where software features can be assembled in weeks, the line between buyer and competitor may be thinner than in previous technology cycles.

Bottom line

Runlayer and Rippling may have buried their lawsuits, but the business lesson remains in plain sight. In AI, a promising enterprise pilot can become the first chapter of a competitive showdown, and the company that looked like a customer yesterday may ship a rival product tomorrow.

For founders, that reality changes everything from sales strategy to product access controls. The legal fight ended without a payout, but the cautionary tale is likely to stick.

The bigger story is not the dismissal itself, but the speed with which AI competition can transform collaboration into rivalry.