A Unicorn Built on AI Agents and Endpoint Paranoia

The Endpoint Is Back

For most of the past decade, security teams fixated on cloud perimeters and SaaS sprawl. Employee laptops remained important but predictable - managed by endpoint detection tools that flagged malicious binaries and suspicious processes. Then generative AI landed on those same devices, and the calculus changed.

Glow, a cybersecurity startup led by former Meta and Snowflake engineering leaders, is betting that shift creates space for a new platform. The company stepped out of stealth this week with $180 million in Series A equity funding at a $1.2 billion post-money valuation. Sequoia Capital, Cyberstarts, Greenoaks, and Redpoint Ventures led the round, joined by Index Ventures, Swish Ventures, Lux Capital, Operator Collective, and Holly Ventures.

At DailyTechWire, we've tracked a wave of AI-native security startups over the past eighteen months, but Glow's premise is more surgical: enterprises now face a combinatorial risk of AI agents, third-party developer tooling, and legacy software all running on the same machines. The startup's founders argue that existing endpoint detection products react to threats after they emerge, whereas the new challenge is to prevent risky code and agents from entering the environment in the first place.

A Team Assembled from Security and Hyperscale

Glow was founded in 2025 by four veterans: Roi Tiger, a former vice president of engineering at Meta; Omer Singer, who led cybersecurity strategy at Snowflake; Ophir Arie, previously vice president of research and development at Claroty; and Arnon Joseph, another engineering leader from Meta. The startup also brought in Emily Heath as chief operating officer. Heath was previously chief information security officer at United Airlines and Docusign, served on the board of Wiz through its $32 billion acquisition by Google, and worked as a partner at Cyberstarts.

That pedigree matters in a market where trust is currency. Enterprise security buyers remain skeptical of point solutions that promise to solve novel problems without demonstrated operational scale. Glow's executive roster gives it credibility with CISOs who remember the hyperscale infrastructure challenges Meta and Snowflake navigated, and the operational discipline United Airlines and Docusign demanded.

What Glow Actually Does

The platform monitors and controls software, AI agents, and developer tools running on employee devices. Under the hood, Glow deploys its own AI agents to continuously map enterprise environments, assess risk in real time, and enforce security policies. According to Tiger, the system has already intercepted malicious npm packages - third-party components used to build applications - before installation, identified AI agents attempting to pull in risky dependencies, and flagged employee devices where endpoint detection and response tools were missing or running in degraded mode.

Glow uses models from Anthropic and Google's Gemini, accessed through Amazon Bedrock, and layers its own software on top to inject enterprise context and improve reliability for security tasks. That architectural choice reflects a broader trend among AI-native security vendors: rather than training foundation models from scratch, they fine-tune or orchestrate existing models with proprietary data and domain logic.

The startup's typical deployment spans tens of thousands of devices across global organizations. Tiger said paying customers already include enterprises in healthcare, retail, and financial services, though the company declined to name them or disclose customer counts. That reticence is standard for stealth-exit startups, but it also makes it harder to assess product-market fit beyond the funding signal.

The Anthropic Shadow and the Mythos Debate

Glow's emergence coincides with rising anxiety over AI-assisted cyberattacks. Anthropic recently unveiled Mythos, a model the company said demonstrated advanced capabilities in identifying and exploiting software vulnerabilities. The disclosure triggered a heated debate in security circles: some researchers argued it represented a step-function increase in attacker productivity, while others pointed out that vulnerability discovery has been partially automated for years.

Regardless of where one lands on Mythos, the perception of escalating AI-driven threats is real among CISOs. That perception creates budget headroom for platforms that promise to counter AI with AI. Glow's pitch leans into that narrative: if attackers use generative models to automate phishing, malware development, and reconnaissance, defenders need agents that can map risk and enforce policy at machine speed.

A Crowded Market and a Contested Category

Glow enters a market dominated by CrowdStrike, Microsoft, SentinelOne, and Palo Alto Networks. Each of those incumbents has invested heavily in AI-powered detection over the past two years. CrowdStrike's Falcon platform now includes generative AI features for threat hunting; Microsoft integrates Copilot into Defender; SentinelOne acquired several AI startups to bolster its Purple AI assistant.

Tiger's counterargument is that those products remain reactive. They excel at detecting malicious behavior after it occurs, but they were not designed to prevent risky AI agents or third-party tooling from entering the environment in the first place. Whether that distinction resonates with buyers depends on how enterprises conceptualize the endpoint in the AI era: as a threat-detection boundary or as an access-control surface.

It is also unclear whether AI-native endpoint security will coalesce into a distinct category or simply become a feature set that incumbents absorb. The funding rounds we've followed across the region suggest investors believe the former, but enterprise buying behavior often favors consolidation over best-of-breed point solutions.

Scale and Headcount in Two Geographies

Glow employs nearly 100 people, roughly 70 percent in Israel and the remainder in the United States. That split is typical for Israeli cybersecurity startups with dual headquarters, and it reflects the concentration of engineering and research talent in Tel Aviv and the surrounding corridor.

The company's rapid scaling - from founding in 2025 to unicorn valuation in under two years - mirrors the trajectory of Wiz, which reached a $10 billion valuation in a similar timeframe before its acquisition by Google. Both companies benefited from founder networks, early customer traction in regulated industries, and a market environment in which large enterprises are willing to pilot new platforms if the leadership team has hyperscale pedigree.

The Valuation Question and Revenue Opacity

Glow joins a cohort of cybersecurity unicorns that achieved billion-dollar valuations before disclosing revenue metrics. That pattern is less common in Asia, where venture investors typically demand proof of recurring revenue before writing nine-figure checks, but it remains standard in Silicon Valley for infrastructure and security startups with strong founder teams and early enterprise traction.

The $1.2 billion valuation implies investor confidence that Glow can capture a meaningful share of the global endpoint security market, which research firms estimate at $20 billion to $25 billion annually. It also reflects the premium investors assign to AI-native platforms in a cycle where generative models are perceived as both a threat vector and a defensive tool.

Whether that confidence translates into durable revenue growth depends on factors Glow has not yet disclosed: customer retention, average contract value, sales cycle length, and competitive win rates against incumbents. Those metrics will become visible if the company raises a Series B or pursues an IPO, but for now the valuation remains a bet on potential rather than a validation of proven economics.

What Comes Next

In the near term, Glow faces two challenges. First, it must prove that AI-native endpoint security is a category rather than a feature. That requires building differentiation that incumbents cannot easily replicate through acquisition or internal development. Second, it must scale go-to-market operations fast enough to justify the valuation while maintaining the product velocity that early customers expect.

The startup's ability to navigate those challenges will depend in part on how quickly enterprises adopt AI agents and how visibly those agents create security incidents. If the next twelve months bring a wave of high-profile breaches tied to compromised developer tools or rogue AI agents, Glow's thesis strengthens. If the threat remains theoretical, the company will need to sell on operational efficiency and risk reduction rather than fear.

At DailyTechWire, we see Glow as a test case for a broader question: can startups build defensible businesses on top of foundation models they do not control, or will the model providers and incumbent platform vendors capture most of the value? The answer will shape not just endpoint security but the entire AI-native software stack over the next several years.