Oak

**Product and the concrete problem it solves.** Oak attacks one of the most persistent and consequential failure modes in enterprise security: no organization can reliably answer, at any given moment, the question "who — and what — has access to which systems, and are they actually using it?" Identity has quietly become the dominant attack surface. Compromised credentials, over-provisioned permissions, orphaned accounts, and stale entitlements are among the most common root causes of breaches, and the legacy identity-governance-and-administration (IGA) and identity-and-access-management (IAM) stack was architected for a slower, human-centric, on-premises world. Oak's answer is what it markets as an **AI-native "Identity Operating System"** — a single control plane that consolidates the fragmented tangle of legacy identity tools and continuously governs the full lifecycle of every identity. Crucially, Oak scopes "identity" to include not just employees and contractors but the exploding population of *non-human* identities: service accounts, machines, workloads, and — increasingly — autonomous AI agents that now act inside enterprise systems alongside people. The company's leadership frames the category shift explicitly, drawing the analogy to how CNAPP consolidated the fragmented cloud-security stack a few years earlier.

**Core technology and how it actually works.** Oak's technical thesis is that identity governance should be evidence-based and continuous rather than record-based and periodic. Instead of trusting static entitlement records or waiting for quarterly access-review campaigns, Oak's platform ingests raw operational data across an environment and builds a **live identity graph** that maps each identity's granted permissions against its *actual* usage. An **AI connector framework** is used to reach across on-premises systems, cloud infrastructure, SaaS applications, and custom/home-grown apps, and Oak claims this AI-driven approach lets it build integrations far faster than traditional connector engineering allows — a meaningful practical advantage in the notoriously integration-heavy identity market. On top of the graph, the system makes **real-time, risk-based decisions and remediations**: it can flag and strip unnecessary or unused permissions as they drift, and it reacts to anomalous signals (for example, logins from unusual locations) as risk events rather than waiting for a scheduled review. The design goal is a shift from operations-based governance (tickets, campaigns, manual certifications) to risk-based, always-on governance — and explicitly to extend that governance model to AI agents, which Oak's founders argue legacy tooling was never built to handle.

**Market, customers, and go-to-market.** Oak sells into the large and structurally growing identity-security market, competing for budget with entrenched IGA suites, IAM/identity providers, and the newer wave of machine- and non-human-identity specialists. The timing thesis is that AI agents are about to multiply the number of identities inside every enterprise by an order of magnitude while simultaneously making credential-based attacks easier to execute, forcing a re-platforming of identity governance that incumbents built for the cloud era are poorly positioned to serve. Oak's go-to-market is enterprise-direct and, notably, launched with the product **already generally available and deployed with paying enterprise customers** — it did not emerge from stealth as a concept but as a shipping platform. Management spent the stealth period consulting with more than 100 CISOs and IAM leaders to shape the product, and the company is orienting much of its commercial motion toward the U.S. market, where most of its staff will soon be based; a Black Hat USA 2026 showcase (Booth 4203) signals an aggressive enterprise-security-buyer launch. Specific customer names have not been disclosed, which is normal at this stage but limits independent verification of traction.

**Traction, funding, and third-party validation.** Oak emerged from stealth on **July 15, 2026** with **$60 million in seed funding** — one of the largest seed rounds ever raised by an Israeli cybersecurity company, and an outlier for a "seed" label. The round was **co-led by Accel, CRV, and Greylock Partners**, with participation from **Hetz Ventures, AlphaDrive Ventures, and strategic angel investors**. The investor roster is itself a strong validation signal: three top-tier funds co-leading a first institutional round is unusual and reflects conviction in the founder and category. There is also a relationship history — Accel previously led the Series A of Morag's prior company, Ermetic, and reportedly held a standing informal commitment to back his next venture. The combination of a large round, tier-one multi-lead syndicate, a generally-available product at launch, and prior-relationship investor conviction constitutes meaningful third-party diligence, even though revenue, customer counts, and unit economics remain undisclosed.

**Founders and team background.** Oak is co-founded by **Shai Morag (CEO)** and **Tal Marom (Chief Product Officer)**, and the founder pedigree is the single strongest asset in the story. Morag is a serial cybersecurity entrepreneur with more than two decades in the field and **three prior exits reportedly totaling roughly $500 million**: Integrity-Project (acquired by Mellanox, now part of NVIDIA, in 2014), Secdo (acquired by Palo Alto Networks in 2018), and Ermetic — a cloud-identity/CIEM company — acquired by Tenable in 2023 for a reported ~$265 million, after which Morag served as Tenable's Chief Product Officer. That last data point matters: Oak is Morag returning to the identity problem with direct, recent operating experience at the scale of a public security vendor. Marom previously led product teams at Tenable and Salesforce and brings Israeli military technical background. The company built a **~50-person team** during stealth, split across Israel and San Francisco, and is hiring aggressively in the U.S. The principal team question is not capability but focus and durability — Morag has publicly framed Oak as his "last company," which is motivating but also raises normal key-person considerations.

**Competitive dynamics.** Oak enters a crowded, well-capitalized field and its differentiation rests on the AI-native, consolidation, and non-human-identity angles rather than on inventing a new category from scratch. (1) Against legacy **IGA incumbents** (SailPoint, Saviynt) it competes on continuous, usage-based governance versus periodic certification campaigns, and on AI-accelerated integration. (2) Against **IAM/identity providers** (Okta, Microsoft Entra ID, Ping) it is complementary in places but competes for the "single control plane" narrative. (3) Against **privileged-access and identity-security** players (CyberArk, and detection-oriented ITDR vendors like Silverfort) it overlaps on risk and least-privilege enforcement. (4) The most strategically important battleground is **machine and AI-agent identity**, where a fast-growing cohort of specialists — Astrix Security, Oasis Security, Token Security, Aembit — is racing for the same "identity for the AI era" position; Oak's own investor acknowledged it will face "plenty of competitors" leveraging AI, making speed-to-scale decisive. Oak's plausible edges are: (i) a genuinely consolidated control plane rather than a point tool; (ii) AI-native connector-building that compresses the integration bottleneck that historically slows identity deployments; (iii) a live, usage-aware identity graph enabling real-time remediation; and (iv) an exceptionally credible founder who has sold into and operated this exact market. The countervailing risk is structural: identity is a sticky, lock-in-heavy market where displacing incumbents is slow, and "AI-native" is a claim many competitors are making simultaneously.

**Defense, security, and resilience dual-use relevance.** Oak's dual-use relevance should be read as a strong *adjacency*, not a fielded defense capability. Identity governance is foundational cyber-resilience infrastructure: government agencies, defense primes, and critical-infrastructure operators are precisely the organizations under the most acute pressure to implement zero-trust architectures, where the animating principle — "never trust, always verify," enforced per-identity and per-access — is exactly the problem Oak automates. Continuous, usage-based least-privilege enforcement, real-time revocation of drifted permissions, and a live map of who and what can touch which systems map directly onto zero-trust mandates (for example, the kind embodied in U.S. federal and defense zero-trust strategies) and onto the security of operational-technology and critical-infrastructure environments. The frontier element — governing **AI-agent and machine identities** — is directly relevant to the secure deployment of autonomous and AI-enabled systems in defense and national-security contexts, where non-human actors increasingly hold privileged access. The honest calibration: Oak is a commercial enterprise-cyber company whose technology is highly relevant to government/defense/critical-infrastructure resilience, but it is not marketed with disclosed defense contracts, certifications (e.g., FedRAMP), or fielded national-security deployments. Its dual-use weight is real and will scale with any public-sector, defense, or critical-infrastructure adoption and accreditation it converts.

**Growth stage, trajectory, and key diligence risks.** Oak reads as an **early-stage but unusually well-capitalized** company: a just-emerged (July 2026) seed-labeled round of extraordinary size, a shipping GA product, a tier-one syndicate, and a proven founder — a profile that behaves more like a Series A than a typical seed. The trajectory bet is that identity re-platforming driven by AI agents creates a large, urgent budget line that a consolidated, AI-native control plane can capture faster than incumbents can retrofit. The key diligence risks are: (1) **intense, well-funded competition** across legacy IGA, IAM, and the AI-agent-identity cohort, in a market where "AI-native" is a widely shared claim; (2) **structural stickiness and lock-in** in identity that make incumbent displacement slow and reference-driven; (3) **undisclosed traction** — no public customer names, revenue, retention, or unit economics, so real adoption depth is unverified; (4) **execution and scaling risk** as the company builds out a U.S.-centric enterprise sales motion at speed; (5) **key-person dependence** on Morag; and (6) on the strategic axis, the **absence of documented public-sector/defense adoption or accreditation** that would convert dual-use adjacency into fielded strategic value. Progression from here would be evidenced by named enterprise (and ideally government/critical-infrastructure) customers, disclosed ARR/retention, credible AI-agent-identity deployments at scale, and any zero-trust/FedRAMP-style accreditation.