Guy Halfon

Automating the part of security nobody wants to do

Guy Halfon spends his days on a problem most people never think about and most companies quietly dread: the sprawling web of vendors, suppliers and partners that every modern business depends on, and the fact that any one of them can become a way in.

As co-founder and CEO of Rescana, Halfon runs a company built around a simple, unglamorous premise. Enterprises now rely on hundreds or thousands of third parties, and each one carries cyber risk. Tracking that risk has traditionally meant spreadsheets, email chains and long security questionnaires that are out of date the moment they are filed. Rescana's pitch is to replace that manual, error-prone routine with autonomous AI agents that discover vendors, classify them, assess their risk and chase down the fixes, end to end and at scale.

The company describes its platform as a set of specialized agents working a single pipeline. A Discovery and Classification agent finds vendors through identity platforms, procurement systems and open-source intelligence. A Risk Assessment agent gathers documentation and produces auditable risk scores. A Communication and Remediation agent handles the outreach and escalation that usually eats an analyst's week. And a Manager agent sits on top, orchestrating policy, reporting and the human-in-the-loop controls that regulated industries insist on. It is a division of labor that mirrors how a real security team already works, only faster and without the burnout.

That framing matters because Rescana's customers are not experimental. The company points its work at banking and capital markets, telecommunications and critical infrastructure, and asset-heavy enterprises - the kinds of organizations that answer to auditors and map their controls to frameworks like NIST CSF, ISO 27001, SOC 2 and MITRE ATT&CK. For them, "the AI did it" is not an acceptable answer. Rescana leans on the idea that every automated decision leaves an auditable trail and a person who can override it.

A practitioner before a founder

Halfon did not arrive at this from a business school case study. He came up through the security trenches. At Playtech, the global gaming and multimedia software operator, he rose from senior cyber security analyst to global manager of information security, running the function that watched over networks operating at enormous scale. Defending an environment like that teaches a specific lesson: risk moves faster than the people assigned to chase it, and the weakest link is often something you do not directly control.

Before Rescana he also founded NoX90, a boutique consultancy delivering high-end security work to large global companies, and served as an advisor to Fortscale, a user-behavior analytics firm later acquired by RSA Security. His public profile carries the handle "zecurity," a small tell about where his identity sits. When he co-founded Rescana in 2016, alongside colleagues and family including Yuval Halfon and Kobi Lechner, it was less a pivot than a continuation - taking the frustrations he had lived through and turning them into a product.

Vega and the plain-English idea

In 2023 Rescana unveiled Vega, an AI-driven, no-code platform for cyber data teams, in closed beta. The core idea is disarmingly practical. Most security tools require an expert to operate them, which creates bottlenecks whenever a business question needs a fast risk answer. Vega uses AI and natural language processing so that someone who has never touched a particular tool can still build collectors, run multi-step analyses, and generate dashboards, reports and alerts by describing what they want. The platform can even read text and decode images as part of its pipeline.

The through-line from Vega to Rescana's current agent-based platform is consistency, not reinvention. Both are answers to the same complaint: security teams do not have a talent problem so much as a time problem. There are only so many hours, and too many of them go to copying data between systems and reformatting the same answer for a new audience. Halfon's bet is that AI can absorb that grind and hand people back the part of the job that actually requires judgment.

Recognition and reach

The approach has drawn notice. In 2023 Rescana was named a CISO Choice Award winner for Best Use of AI, an award judged by the people who would actually buy the product. The following year the company announced a partnership with cloud security firm Wiz and the consultancy Aspiration Japan to deliver a cloud security offering tailored to the Japanese market - a signal that a small Tel Aviv and New York operation was reaching well beyond its home turf.

What stands out about Halfon is the market he chose. Third-party risk management is not the flashy corner of cybersecurity. It does not produce dramatic breach headlines or slick threat-hunting demos. It is the plumbing - and, as any security veteran will tell you, the plumbing is usually where the real leaks happen. Choosing to build there is a practitioner's instinct: go where the pain is boring, repetitive and expensive, because that is where automation earns its keep.

Rescana frames its mission as an effort to "shift the balance," letting security, legal and risk teams scale their oversight of vendors without adding headcount. It is a modest-sounding goal that hides an ambitious one. If a handful of agents really can carry the weight of a whole risk program, the economics of security change for every enterprise that has quietly given up on tracking its suppliers. Whether that fully arrives is still being written. But Halfon has spent his career on exactly this blind spot, and he is not treating it as a side project.