Here's the uncomfortable truth about AI-generated code. It ships fast, it compiles clean, and nobody's entirely sure what it does until it's already in production. Unleash, the Oslo-based open-source feature management company, just raised $35 million to make sure that when something goes wrong, and it will, you can turn it off before it takes down the system. Not tomorrow. Not after a war room. Instantly.

The Series B was led by One Peak, with continued backing from Spark Capital, Frontline Ventures, and Firstminute Capital. For a company built on the un-glamorous work of feature flags and runtime governance, it's a significant vote of confidence in a category that's suddenly urgent.

Because here's what's changed. DORA research shows that as AI adoption rises in engineering teams, software stability declines. Code is being written faster than it's being understood. And the people who get called at 2 AM when something breaks? Still human. Still tired. Still trying to figure out which of the seventeen features deployed that afternoon caused the spike in error rates.

In 2025, both Google Cloud and Cloudflare suffered major outages caused by feature rollout failures. Not security breaches. Not hardware problems. Features that went live without proper governance. Egil Osthus, Unleash's CEO and co-founder, points to these incidents as evidence that the problem isn't niche. It's structural. If Google can't manage feature rollouts safely, what chance does a 500-person company have?

DevOps transformed how code gets into production. Continuous integration, continuous deployment, infrastructure as code. The tooling is mature. But once that code is running, controlling what actually turns on, who sees it, and how quickly it can be reversed? That's a different problem entirely. It sits in the gap between deployment and user experience, and until recently, most companies addressed it with hope and manual runbooks.

Unleash calls its category FeatureOps, and the pitch is straightforward: give engineering teams runtime control over every feature, with audit trails, kill switches, and compliance controls built in. For enterprises deploying AI-generated code at scale, this isn't optional anymore. It's the difference between moving fast and moving fast into a wall.

Unleash started as an open-source project, and that decision has shaped everything about how the company grows. The open-source version is one of the most popular feature management tools on GitHub. It's the way most developers first encounter Unleash: they find it while looking for a self-hosted alternative to LaunchDarkly, tinker with it during a hackathon, integrate it into a side project, and then bring it to their employer when the employer starts caring about feature governance.

That bottom-up adoption is the company's moat. By the time Unleash's sales team reaches out to an enterprise prospect, there's often already a team inside the company using the open-source version. The conversation shifts from 'why do we need feature management' to 'how do we upgrade what we already use.' It's a vastly easier sale.

The commercial layer adds the governance features that enterprises demand: role-based access controls, advanced audit logging, compliance reporting, and deep integration with CI/CD pipelines. It's the classic open-source business playbook, executed by a team that understands the developer audience deeply because they've been serving it for years.

Alongside the funding announcement, Unleash launched Impact Metrics, a new product that lets teams measure the actual effect of a feature rollout in real time. Not just whether it deployed successfully, but whether it moved the metrics it was supposed to move. Did the new checkout flow increase conversion? Did the updated recommendation algorithm improve engagement? Or did it silently degrade performance in ways that won't show up until next quarter's numbers?

This is the piece that most feature management tools miss. They tell you a flag is on or off. They don't tell you whether the feature behind that flag is helping or hurting. For AI-generated features, where the developer might not fully understand the code's behavior, that feedback loop is critical. Without it, you're flying blind with the autopilot on.

Osthus describes Impact Metrics as closing the governance loop. 'You need to know what shipped, who approved it, what it did, and whether you should keep it.' Simple framework. Gets complicated fast when you're shipping dozens of features a day across multiple teams, each with their own metrics and rollout schedules.

Norway isn't the first country you'd associate with open-source enterprise software. But Unleash joins a small, credible cohort of Norwegian companies that have found commercial traction with open-source products without relocating to San Francisco. The BusinessWire announcement emphasized the company's Oslo roots, and One Peak leading the round signals European investor confidence in the open-source enterprise model that has historically been dominated by US firms.

The challenge ahead is familiar to any open-source company. Convert free users to paying customers. Expand into regulated industries where governance features aren't optional but mandatory. Build a sales team that can compete with American incumbents, particularly LaunchDarkly, which has raised over $300 million and established itself as the market leader. The $35 million gives Unleash runway to attempt all three simultaneously.

For developers who've been using the open-source version for years, the question is whether enterprise governance features are worth paying for. For their CISOs and compliance officers, it's not even a question. It's a requirement. And as AI-generated code becomes the norm rather than the exception, the audience that considers feature governance 'nice to have' is shrinking rapidly. Unleash's timing might be perfect.