Koi

Koi is an AI-powered endpoint security tool that helps organizations understand exactly what software does before it runs. It scans browser extensions, npm packages, and other marketplace software to uncover hidden risks—like malicious code, data exfiltration, or backdoors. Instead of relying on simple reputation lists, Koi uses large language models to read the actual code of non-binary software and compare it against what the publisher claims. This gives security teams a clear, data-driven risk score for every piece of software in their environment.

How does Koi work? It starts by proactively scanning app stores, registries, and marketplaces to find new software. Then it assesses the publisher—looking at their online presence, region, and cross-marketplace reputation. The real magic happens when Koi's LLMs analyze the actual code to see what the software was programmed to do, versus what it says it does. It even runs the software in a sandbox to capture real network and endpoint activity. All this data feeds into a risk score that updates automatically when new versions are released. AI agents continuously enrich the software profile with breach intelligence, vulnerability data, licensing info, and capability insights.

Koi is built for security, IT, GRC, and SOC teams that need to govern software at scale. It's especially useful for enterprises that rely on open-source packages, browser extensions, and third-party add-ons—where supply chain attacks are common. Instead of guessing or trusting manual reviews, Koi gives you a clear, automated risk assessment for every install. If you've ever worried about what a browser extension is really doing, Koi offers a concrete answer.