Ray Security

• Data-security incumbents such as Microsoft Purview and Varonis, or better-funded peers like Cyera and BigID, could add predictive AI-agent access controls and erase Ray's differentiation before it wins scale.
• Ray's predictive engine is unproven in public. If early deployments show it cannot reliably predict access without heavy tuning, the core pitch collapses to conventional data-security posture management.
• Ray discloses one named reference and a press claim of Fortune 500 customers but no deployment scale or contract metrics. If it cannot convert early references into disclosed, referenceable scale, the go-to-market thesis stays unproven.
• Ray holds no public security attestation. Regulated buyers in finance and healthcare may decline production deployments until SOC 2 or ISO 27001 is in place.
• In-tenant learning yields no cross-customer data asset, so Ray gains no compounding data advantage that would slow a well-funded rival.

Ray Security addresses a gap created as AI agents began reaching enterprise data alongside employees and systems. Ray argues that autonomous agents and copilots now reach the same corporate data stores that people use, often with little oversight, and that static controls built for known users and fixed permissions cannot govern access that changes by the hour.

Ray aims the pitch at security teams in data-heavy, regulated sectors such as finance, healthcare, and technology. The problem it describes is real and widely discussed, but its specific framing, including a claim to cut data risk by 90 percent, comes from Ray itself. No independent study quantifies the exposure the way Ray states it, so the problem is clearly defined but not independently measured. [s5, s6, s1]

Ray Security sells one platform that pairs data discovery with usage-based protection. Its predictive engine analyzes historical access patterns, data behavior, and business cycles to forecast which data will be accessed, separate business-critical from dormant data, and flag anomalous AI-agent behavior. On top of that, dynamic protection blocks unauthorized activity, applies automated controls, and restricts what AI agents can reach, retrieve, and relay.

The AI-agent focus runs through the product. Ray discovers shadow-AI tools and agents inside and outside the network, maps what each one accessed, and enforces boundaries so copilots and agents cannot surface data outside policy. The capability descriptions are specific, but they live only on Ray's own marketing pages. No public technical documentation, public demo, open-source code, or third-party evaluation appears in the reviewed sources, so the depth behind the predictive claim stays unverified. [s1, s2]

Ray competes in data security, one of the most crowded parts of the market, and stakes its position on prediction rather than after-the-fact scanning. Established data-security incumbents and data posture-management vendors address the same enterprise buyer, most already sell data discovery and classification, and several are positioned to add controls for how AI agents reach data.

Ray's sharper positioning is the AI-agent angle. The AI Defense Matrix catalog lists Ray Security for agentless discovery of AI agents and shadow-AI tools that reach enterprise data, mapping what each accessed and governing their access. That framing separates Ray from pure data-scanning tools today, but it is also the direction larger vendors are moving, so the distance is unlikely to hold without proof that the predictive engine works. [s1, s8, s6]

Ray's public traction is early but no longer empty. Its homepage carries a customer testimonial from Erik Hart, the CISO of Cushman & Wakefield, describing a delivered deployment, and Calcalist reports that Ray has already secured Fortune 500 customers in finance, healthcare, and technology. Partner-oriented testimonials from firms such as MOBIA and Compuquip appear on the homepage, alongside a direct sales motion with Free Trial and Contact Us calls to action but no published pricing.

What Ray does not disclose is scale. It publishes no customer count, revenue, or deployment metrics, so the traction reads as credible early proof rather than demonstrated breadth. The supporting signals beyond customers are an $11 million seed round co-led by Venture Guides and Ibex Investors and a 2026 World Economic Forum Technology Pioneer selection. [s1, s6, s10, s4, s7]

Ray Security was founded in September 2024 by Ariel Zamir as chief executive, Eric Wolf as chief business officer, and Dekel Levkovich as chief technology officer. Independent reporting describes the three as veterans of NetApp, Check Point, Guardicore, and Axonius, which is verifiable senior experience at well-known security and infrastructure companies.

That pedigree is a genuine asset for a seed-stage company, and it supports the technical ambition of a predictive data platform. The public record does not document a prior founder-led exit or a sustained research and publication record for the team, so the background is strong without reaching the standout tier. [s5, s6]

Ray Security shows no public compliance attestation, which is common a year past founding but material to the regulated buyers it targets. A probe on 2026-07-04 of Ray's trust and security subdomains, its /security, /trust, and /compliance paths, and its homepage found no SOC 2, ISO 27001, or other first-party certification. The homepage references HIPAA and GDPR only as audit requirements the platform helps customers meet, not as Ray's own attestations.

For a data-security vendor asking enterprises to route sensitive-data governance through its platform, a visible attestation is table stakes that Ray has not yet published. Buyers in finance and healthcare will likely require SOC 2 Type II or the equivalent before production use. [s9, s1]

Add analyzed competitors to compare them side by side with Ray Security.

Ray Security targets mid-size to large enterprises in data-heavy, regulated industries, naming finance, healthcare, and technology as priority sectors. The buyer is the enterprise security team responsible for data protection and, increasingly, for governing how AI tools touch that data. Independent reporting frames Ray's plan as increasing its footprint among these high-risk, data-intensive organizations.

The entry wedge is narrow inside a broad market. Data security spans discovery, classification, loss prevention, and posture management, and Ray's marketing leads with the AI-agent access angle even as its platform page presents broader inventory, classification, and remediation capabilities. The segmentation is coherent, and press coverage reports Fortune 500 customers in finance, healthcare, and technology, early conversions in the named sectors, though without named accounts or deployment depth.

Ray's central capability is prediction applied to data access. The platform analyzes historical access patterns, data behavior, and business cycles to forecast which data will be needed, distinguish business-critical from dormant data, and apply protection before access rather than after a scan. Ray positions this as a shift from static safeguards to protection that tracks real usage.

The AI-agent capabilities extend the same engine. Ray discovers shadow-AI tools and agents, maps what data each accessed, maintains data lineage across AI interactions, and enforces boundaries on what agents can reach, retrieve, and relay. It also markets agentless data loss prevention that controls what large language models and agents can pull.

AI is the method throughout rather than a bolted-on feature, and that is the genuine strength of the design. The caveat is proof. Every capability sits on Ray's own pages with no external validation, and the predictive claim, the part hardest to build, is also the part with no public evidence behind it.

Ray sells through a direct enterprise motion. The website offers Free Trial and Contact Us calls to action but no published pricing, which fits a product that instruments an organization's data and usually signals negotiated enterprise deals.

Early traction is real but thinly disclosed. Ray's homepage carries a Cushman & Wakefield CISO testimonial describing a delivered solution and swift integration, Calcalist reports it has secured Fortune 500 customers in finance, healthcare, and technology, and representatives of MOBIA and Compuquip endorse the platform on its homepage. Ray also raised $11 million in seed funding co-led by Venture Guides and Ibex Investors and was named a 2026 World Economic Forum Technology Pioneer. It publishes no customer count, revenue, or deployment metrics, so the motion shows credible early proof without demonstrated scale.

Ray publishes no pricing. The site offers a Free Trial and Contact Us calls to action but no price or tier, which points to negotiated enterprise contracts rather than transparent or usage-based pricing. For a platform that scales with the volume and sensitivity of a customer's data, that is a conventional choice at this stage.

Because the company has not disclosed a pricing unit, it is not yet possible to say what Ray believes buyers pay for, whether by data volume, users, monitored environments, or protected data stores. The pricing story is therefore undefined in public, which is normal at seed stage but leaves budget-sizing questions open for buyers.

Ray delivers as a SaaS platform and emphasizes agentless discovery. The AI Defense Matrix catalog describes Ray's shadow-AI discovery as agentless, meaning no agent is required on the data source. That lowers the adoption barrier and speeds time to first value.

The platform's operational promise is real-time reaction rather than alert-and-wait. Ray describes applying automated protection across environments and blocking unauthorized activity as it happens, replacing the older loop of scanning, alerting, and manual triage. How well that automation performs in production is hard to judge from outside: the homepage carries a Cushman & Wakefield account of delivery and integration, and no third-party test is public.

Ray has not published a security attestation, and for a data-security vendor that is a live gap rather than a formality. A probe on 2026-07-04 of Ray's trust and security subdomains, its /security, /trust, and /compliance paths, and its homepage found no SOC 2, ISO 27001, or comparable first-party certification. The homepage mentions HIPAA and GDPR only as audit requirements the platform helps customers satisfy.

Trust matters doubly for Ray because its value proposition asks a customer to let an outside platform observe and govern access to its most sensitive data. Many enterprise buyers in finance and healthcare may require SOC 2 Type II or ISO 27001 before broad production use, so publishing an attestation is likely a near-term gate on the mid-to-large deals Ray is chasing.

Ray presents itself as a single data-security platform and is beginning to build a channel around it. Its integration surface is the customer's own data stores and the AI tools that touch them, it says it integrates with cloud, on-premises, and hybrid sources, with its shadow-AI discovery described as agentless, and it carries homepage testimonials from firms such as MOBIA and Compuquip alongside Partners and Become a Partner navigation. The company has not published a named integration catalog or connector list.

The platform's reach into AI tools, copilots, and agents is where a deeper ecosystem would form, since governing agent access requires visibility across many AI endpoints. That breadth and the hybrid cloud and on-premises coverage are asserted on the marketing pages and in press, but not yet documented with named integrations, so the ecosystem position is early rather than an evidenced strength.

Ray Security's founding team combines security and infrastructure backgrounds. Ariel Zamir leads as chief executive, Eric Wolf as chief business officer, and Dekel Levkovich as chief technology officer, and independent reporting describes the three as veterans of NetApp, Check Point, Guardicore, and Axonius. Those are established names in security and data infrastructure, which lends credibility to an ambitious technical claim.

What the public record does not yet show is a prior founder-led exit or a sustained body of published research from this team. The pedigree suggests execution risk is lower than average for a seed company, but it is employment history at notable firms rather than a track record of building and exiting a company of their own.