Echo raises $35M to safe the enterprise cloud’s base layer — container pictures — with autonomous AI brokers

As enterprises speed up the deployment of LLMs and agentic workflows, they are hitting a vital infrastructure bottleneck: the container base pictures powering these purposes are riddled with inherited safety debt.

Echo, an Israeli startup, is asserting a $35 million in Sequence A funding as we speak (bringing its to-date complete to $50 million in funding) to repair this by basically reimagining how cloud infrastructure is constructed.

The spherical was led by N47, with participation from Notable Capital, Hyperwise Ventures, and SentinelOne. However the actual story is not the capital—it is the firm’s formidable purpose to substitute the chaotic open-source provide chain with a managed, “secure-by-design” working system.

The Hidden Working System of the Cloud

To know why Echo issues, you first have to perceive the invisible basis of the trendy web: container base pictures.

Consider a “container” like a transport field for software program. It holds the utility code (what the builders write) and every part that code wants to run (the “base picture”). For a non-technical viewers, the greatest method to perceive a base picture is to examine it to a brand-new laptop computer. Once you purchase a pc, it comes with an Working System (OS) like Home windows or macOS pre-installed to deal with the fundamentals—speaking to the exhausting drive, connecting to Wi-Fi, and working applications. With out it, the laptop is ineffective.

In the cloud, the base picture is that Working System. Whether or not an organization like Netflix or Uber is constructing a easy net app or a fancy community of autonomous AI brokers, they rely on these pre-built layers (like Alpine, Python, or Node.js) to outline the underlying runtimes and dependencies.

Right here is the place the threat begins. In contrast to Home windows or macOS, which are maintained by tech giants, most base pictures are open-source and created by communities of volunteers. As a result of they are designed to be helpful to everybody, they are typically full of “bloat”—a whole bunch of additional instruments and settings that the majority corporations do not really need.

Eylam Milner, Echo’s CTO, makes use of a stark analogy to clarify why this is harmful: “Taking software program simply from the open supply world, it is like taking a pc discovered on the sidewalk and plugging it into your [network].”

Historically, corporations strive to repair this by downloading the picture, scanning it for bugs, and trying to “patch” the holes. Nevertheless it is a shedding battle. Echo’s analysis signifies that official Docker pictures typically comprise over 1,000 identified vulnerabilities (CVEs) the second they are downloaded. For enterprise safety groups, this creates an unattainable sport of “whac-a-mole,” inheriting infrastructure debt before their engineers write a single line of code.

The “Enterprise Linux” Second for AI

For Eilon Elhadad, Echo’s co-founder and CEO, the trade is repeating historical past. “Precisely what’s occurred in the previous… everyone run with Linux, after which they transfer to Enterprise Linux,” Elhadad instructed VentureBeat. Simply as Purple Hat professionalized open-source Linux for the company world, Echo goals to be the “enterprise AI native OS”—a hardened, curated basis for the AI period.

“We see ourselves in the AI native period, the basis of every part,” says Elhadad.

The Tech: A “Software program Compilation Manufacturing unit”

Echo is not a scanning device. It does not search for vulnerabilities after the reality. As a substitute, it operates as a “software program compilation manufacturing facility” that rebuilds pictures from scratch.

In accordance to Milner, Echo’s method to eliminating vulnerabilities depends on a rigorous, two-step engineering course of for each workload:

1. Compilation from Supply: Echo begins with an empty canvas. It does not patch current bloated pictures; it compiles binaries and libraries immediately from supply code. This ensures that solely important parts are included, drastically lowering the assault floor.

2. Hardening & Provenance (SLSA Stage 3): The ensuing pictures are hardened with aggressive safety configurations to make exploitation tough. Crucially, the construct pipeline adheres to SLSA Stage 3 requirements (Provide-chain Ranges for Software program Artifacts), making certain that each artifact is signed, examined, and verifiable.

The end result is a “drop-in alternative.” A developer merely adjustments one line of their Dockerfile to level to Echo’s registry. The applying runs identically, however the underlying OS layer is mathematically cleaner and freed from identified CVEs.

AI Defending Towards AI

The necessity for this stage of hygiene is being pushed by the “AI vs. AI” safety arms race. Unhealthy actors are more and more utilizing AI to compress exploit home windows from weeks down to days. Concurrently, “coding brokers”—AI instruments that autonomously write software program—are changing into the primary turbines of code, typically statistically deciding on outdated or weak libraries from open supply.

To counter this, Echo has constructed a proprietary infrastructure of AI brokers that autonomously handle vulnerability analysis.

• Steady Monitoring: Echo’s brokers monitor the 4,000+ new CVEs added to the Nationwide Vulnerability Database (NVD) month-to-month.

• Unstructured Analysis: Past official databases, these brokers scour unstructured sources like GitHub feedback and developer boards to determine patches before they are extensively printed.

• Self-Therapeutic: When a vulnerability is confirmed, the brokers determine affected pictures, apply the repair, run compatibility exams, and generate a pull request for human overview.

This automation permits Echo’s engineering crew to preserve over 600 safe pictures—a scale that will historically require a whole bunch of safety researchers.

Why It Issues to the CISO

For technical decision-makers, Echo represents a shift from “imply time to remediation” to “zero vulnerabilities by default.”

Dan Garcia, CISO of EDB, famous in a press launch that the platform “saves no less than 235 developer hours per launch” by eliminating the want for engineers to examine false positives or patch base pictures manually.

Echo is already securing manufacturing workloads for main enterprises like UiPath, EDB, and Varonis. As enterprises transfer from containers to agentic workflows, the capability to belief the underlying infrastructure—with out managing it—could also be the defining attribute of the subsequent era of DevSecOps.

Pricing for Echo’s resolution is not publicly listed, however the firm says on its website it costs “primarily based on picture consumption, to guarantee it scales with the way you truly construct and ship software program.”