Back
#246
October 6, 2025

EP246 From Scanners to AI: 25 Years of Vulnerability Management with Qualys CEO Sumedh Thakar

Guest:

Topics:

Vulnerability Management
29:29

Subscribe at Spotify

Subscribe at Apple Podcasts

Subscribe at YouTube

Topics covered:

  • How did vulnerability management (VM) change since Qualys was founded in 1999? What is different about VM today?
  • Can we actually remediate vulnerabilities automatically at scale? Why did this work for you even though many expected it would not?
  • Where does cloud fit into modern vulnerability management?
  • How does AI help vulnerability management today? Is this primarily better prioritization? What is real?
  • What is this Risk Operations Center (ROC) concept and how it helps in vulnerability management?

Do you have something cool to share? Some questions? Let us know:

Transcript

The discussion opened by establishing that Vulnerability Management (VM) remains foundational to cybersecurity, noting that vulnerability exploitation is the second leading cause of breaches, a trend expected to soon surpass credential theft and phishing. This validates VM’s "evergreen" status, a concept one host quipped could be familiar to a security leader from 1995.

The core change is one of scale and speed. The sheer volume of new CVEs is overwhelming security teams. This is a natural consequence, described by the guest as "just physics," resulting from massive digital transformation and the corresponding explosion in written and deployed software.

Crucially, the time taken for attackers to exploit a vulnerability has decreased by approximately 70%. In the past, a "gold star customer" might have scanned every 90 days and had 90 days to patch; today, that timeframe is dangerously compressed. This new reality forces a complete re-evaluation of the VM process, making it too expensive and too fast to maintain a catch-all security posture.

The Shift from CVSS to Prioritized Business Risk

The single most significant point of friction in modern VM is prioritization. Traditional practices, such as attempting to fix all high or critical vulnerabilities based purely on the Common Vulnerability Scoring System (CVSS), are no longer tenable.

The hosts and guest explored the counter-intuitive yet necessary decision to prioritize a lower-scoring vulnerability over a high-scoring one. The guest offered a compelling justification: a CVSS score of 9.8 with no associated exploit is a lower priority than a 6.3 that is actively being used in a kill chain. The challenge lies in providing the security team with the justification to ignore the 9.8—a moment that still causes "squeamishness" among seasoned security professionals.

This required maturity is achieved by injecting two key data points into the prioritization engine:

Threat Intelligence: Data on active exploitation, known exploits, and attacker trends.

Business Context (Value at Risk): The financial impact of an asset being compromised.

The guest offered a classic example: given a choice between fixing a system with a risk score of 900 that supports a business unit generating $2 million annually, or a system with a score of 750 that supports a unit generating $500 million, the latter becomes the immediate priority. Any prioritization conversation without some aspect of Value at Risk is deemed "not helpful."

Operationalizing Remediation: The Value Proposition

A major theme was the candid realization that the "only value that you get in the vulnerability management cycle is if you actually fix it." The process is not about scanning or dashboard selfies; it’s about remediation.

Automated Patching: Defying the Critics

The guest acknowledged the initial industry skepticism—which included one host—toward automated patching. The rationale for this past skepticism centered on "snowflake" servers and the risk of operational fragility ("it's gonna blow up").

The success of automated patching (validated by Qualys's deployment of 110 million patches in one year) is attributed to a risk-based and scoped approach, not an all-or-nothing proposition. Automation is selectively applied to low-risk, high-volume assets (like employee laptops and browsers, referencing Google Chrome’s frequent zero-day fixes) to free up IT teams to focus on mission-critical assets.

The Remediation Buffet: Beyond the Patch

The focus has expanded beyond mere patching to a "remediation buffet" that offers multiple options for closing the risk gap:

Software Mitigation: Applying "smart tweaks" (e.g., changing file permissions, editing a registry key, or shutting down a non-essential service) to prevent an exploit without applying a full patch. This is crucial for zero-day events where a patch is not yet available.

Software Removal: Uninstalling vulnerable software that has not been used on a machine for a long period, which is simple, effective, and often overlooked.

Contextual Risk Acceptance: Incorporating data from other security layers, such as Endpoint Detection and Response (EDR) memory protection, to determine that the risk of exploit is already mitigated, even if the underlying vulnerability technically exists.

The Strategic Horizon: Risk Operations Center and Agentic AI

The ultimate shift is conceptual: moving from Attack Surface Management (which focuses on every door and window) to Risk Surface Management (which focuses on what can be lost and the financial impact).

The Risk Operations Center (ROC)

The guest introduced the concept of the Risk Operations Center (ROC) as a category shift, not just a product. The ROC aims to:

Ingest all finding data from multiple tools (not just the vendor's scanner).

Apply threat intelligence and business context.

Expand scope beyond traditional CVEs to include misconfigurations and identities, recognizing that an unpatched machine with open read-write permissions to the network has a horrible holistic risk posture, regardless of its CVE count.

This integrated approach enables the CISO to provide a unified posture view tied directly to business outcomes, rather than disparate dashboards from ten different products.

The Integration of Agentic AI

Generative and Agentic AI are positioned as essential force multipliers for defenders to counter the volume and speed of attacker automation. AI is not an add-on but is embedded throughout the platform.

Key AI use cases include:

Risk Prediction: Analyzing the attack surface to identify which findings have the highest possibility of a successful exploit.

Mitigation Recommendation: Leveraging AI to suggest mitigations that are least likely to cause operational disruption.

End-to-End Workflow Automation (Agentic AI): The introduction of a Cyber Risk Agent Marketplace where specialized digital agents (e.g., Agent Sarah for Patch Tuesday or Agent Nova for External Attack Surface Management) automate complex, multi-step tasks that would take a human defender weeks to perform. This represents a paradigm shift where the CISO’s team consists of human experts augmented by digital agents.

The CISO's Board-Level Conversation

The final segment focused on the critical task of translating security outcomes into financial and strategic business terms for the Board. The historical problem is that CISOs finally got to the Board, only to realize the conversation was disconnected—the Board didn't understand the CISO's technical metrics, and the CISO didn't understand the Board's financial concerns.

The ROC and CRQ (Cyber Risk Quantification) provide the necessary framework to shift this dialogue:

From: "We reduced open vulnerabilities from 31,000 to 28,000."

To: "Our $500 million business unit is currently at a 75% possibility of incurring a $10 million per day loss over seven days due to a ransomware attack. A $300,000 investment will reduce that possibility by 80%."

The guest highlighted that strategic decisions are then framed around risk mitigation, risk acceptance, and risk transfer (cyber insurance). A partnership with a cyber insurance company was noted, offering premium discounts based on a demonstrably lower risk score derived from the ROC approach.

The ultimate win for the CISO is the ability to show the Board, and the IT team, that out of 65 million total vulnerabilities, the process has identified the 304,000 that genuinely matter to the business, thereby reducing the workload on IT teams by focusing their effort where it provides the maximum risk reduction.

Timeline of Key Conversation Shifts

Introduction & The Status Quo

Podcast welcome, host introductions, and guest introduction (Sumedh Thakar, CEO of Qualys).

Establishing vulnerability management (VM) as an "evergreen" but increasingly urgent problem.

Identification of the core change: Massive increase in vulnerability volume (new CVEs) and attack velocity (speed of exploitation).

Prioritization and Financial Constraints

Recognition that organizations have a fixed budget and cannot fix all vulnerabilities.

Introduction of the prioritization dilemma: Which vulnerabilities should be fixed for maximum risk reduction?

Critique of relying solely on CVSS scoring for prioritization.

The essential argument for a Risk-Based Approach: Combining threat intelligence with business context (Value at Risk).

The Operational Reality of Remediation

The assertion that remediation is the only value derived from the VM process.

Discussion of the initial industry skepticism toward automated patching.

Justification for the success of automated patching by balancing security risk with operational risk (e.g., using automation for low-risk, high-volume assets).

Expansion of remediation beyond patching to a "Remediation Buffet" including mitigation, configuration tweaks, and uninstallation of unused software.

The Strategic Vision: Risk Operations Center

Introduction of the concept of Risk Surface Management as a superior strategy to Attack Surface Management.

Definition of the Risk Operations Center (ROC) as a conceptual framework for holistic risk analysis.

Expansion of the security scope beyond CVEs to include misconfigurations and identities for a true holistic risk posture.

The AI and Agentic Future

The necessity of using AI for defense to counter attackers' use of AI and automation.

Discussion of Generative AI use cases: identifying exploitable findings and recommending optimal mitigations.

Introduction of Agentic AI via a Cyber Risk Agent Marketplace (e.g., Agent Sarah, Agent Nova) for automating end-to-end security workflows.

The CISO-Board Alignment

Focus on the critical need to shift the CISO's board conversation from technical metrics to Cyber Risk Quantification (CRQ).

Framing security investment around the probability of financial loss, risk acceptance, and risk transfer (cyber insurance).

Final strategic tip for CISOs: start the CRQ journey by modeling cyber risk on existing business risk frameworks and aim for directional accuracy over perfection.

View more episodes