HelmGuard Raises $7.3 Million to Build Continuous Risk Assurance With AI Agents

Key Takeaways

• $7.3 Million Seed Round: HelmGuard raised $7.3 million in a round co-led by Infinity Ventures and Frontline, with participation from FinTech Collective, Stage 2 Capital and Entrepreneurs First.
• Direct Risk Verification: HelmGuard uses AI agents to collect and assess evidence from source systems, seeking to move assurance beyond reliance on questionnaires, reports and other static documents.
• AI Agent Assurance: Part of the funding will support an assurance layer designed to evaluate AI-agent behavior at runtime as organizations deploy agents into operational workflows.
• Verified Risk Network: HelmGuard is developing a network that would allow organizations to exchange continuously verified claims while applying their own risk appetites, control frameworks and regulatory requirements.
• U.S. Expansion: The company plans to expand its presence in New York and San Francisco and hire across engineering and go-to-market roles.

Deep Dive

HelmGuard has raised $7.3 million in seed funding to expand its AI-driven approach to governance, risk and compliance, as the company looks to replace some of the questionnaires and periodic document reviews that still underpin corporate risk assurance with agents capable of examining evidence directly.

The round, announced Wednesday, was co-led by Infinity Ventures and Frontline, with participation from FinTech Collective, Stage 2 Capital and Entrepreneurs First. HelmGuard said the funding will support its continued expansion in the U.S., including a presence in New York and San Francisco, along with hiring across engineering and go-to-market roles and further development of an assurance layer designed to evaluate AI-agent behavior at runtime.

The company is building around a problem that has become increasingly difficult to ignore in GRC: an assessment can be thorough and still describe a version of an organization that no longer exists. Security questionnaires, policies, certifications and audit reports remain fixtures of assurance programs, but the systems behind them can change long before the next assessment arrives.

HelmGuard's answer is to move more of the assessment itself closer to those systems. Its AI agents collect and assess risk signals from source evidence, including configurations, logs, attack-surface data and documentation, with the aim of giving security and compliance teams a more current view of controls and exposures. The company says that approach can reduce assessment cycles from weeks to hours.

John Daley, HelmGuard's co-founder and CEO, and Jack Miller, its co-founder and CTO, argue that the document-heavy model was never intended to be the ideal form of assurance. It was a practical compromise. Direct, continuous verification was too expensive, so organizations relied on proxies such as SOC 2 reports, security questionnaires and policy documents that described controls rather than directly observing them. The founders contend that AI agents are beginning to change those economics.

That distinction sits at the center of HelmGuard's pitch. Plenty of GRC platforms are already using AI to read questionnaires, process reports and accelerate existing compliance workflows. HelmGuard is trying to automate more of the underlying verification instead.

"Most compliance platforms were built to document a process, not to reach a conclusion," Daley, who spent eight years as an executive at Palantir, said in the company's announcement. "Now they're using AI to produce those documents faster, which doesn't help anyone decide anything."

The argument has acquired greater urgency as the systems being assessed have begun changing faster. Software vendors are introducing AI capabilities, models are being updated and agents are gaining access to tools and data that can materially alter what a system can do. HelmGuard's founders point to three pressures on traditional assurance in particular: organizations are deploying software capable of reasoning and acting, AI is lowering the cost of vulnerability discovery and exploitation, and software dependencies can now change weekly or even daily.

"An AI vendor's risk profile changes with every model update and every new tool its agents can call," Miller said in the announcement. "As a result, a certification issued months ago describes a reality that our customers cannot rely upon."

HelmGuard's platform brings together risk, security and compliance information and uses specialized AI agents to perform work including third-party risk assessments, control-gap assessments and agent assurance. Instead of relying principally on an organization's answers about its controls, the agents can collect and assess evidence from source systems.

The resulting unit of assurance, in HelmGuard's terminology, is a "verified claim." Each claim can be connected to the controls intended to mitigate the underlying risk, the requirements it implicates and the person or team responsible. HelmGuard says its findings include citations to the evidence examined, a reasoning trail and a confidence score, while customers can establish thresholds based on risk, confidence, criticality or particular control types that require human review.

The distinction matters because speed alone is not much of an achievement if automation simply produces weaker assessments more efficiently. HelmGuard's founders argue that direct examination can instead increase rigor by allowing organizations to move from occasional sampling toward broader and more continuous observation of their environments. In their telling, completing an assessment faster is a consequence of the model, not its purpose.

HelmGuard says organizations in financial services, insurance, healthcare and industrials are already using the platform, with customers across the U.S., Canada, U.K., Hong Kong and South Africa. According to the company, one U.S.-based insurance customer used HelmGuard to conduct a full risk assessment of 1,250 counterparties in less than a week and subsequently migrated from its legacy platform. HelmGuard said its forward-deployed engineers completed that migration in under 10 days.

Another customer, described by HelmGuard as a global telehealth and telecommunications company, has automated its customer assurance process and reduced first-response times from days to minutes. Callosum, a London-based company selling advanced AI technologies into regulated industries, has also used HelmGuard to design and operate its security and compliance program. Those performance figures and customer outcomes were provided by HelmGuard.

The seed funding will also help the company pursue a problem that sits beyond conventional third-party risk: how organizations assure the AI agents they are beginning to deploy themselves.

A policy can establish what an agent is permitted to do, but HelmGuard's premise is that governing an autonomous system also requires observing what it actually does. The company is developing an agent assurance layer intended to evaluate that behavior at runtime, allowing an organization to define a claim about an agent and continuously test whether the evidence supports it.

The founders offered the example of an organization seeking to establish that no production agent simultaneously has access to sensitive data, exposure to untrusted content and the ability to send information externally. HelmGuard's platform could continuously verify that claim against agent traces, with the claim and each verification becoming part of the record.

That approach would also change the audit trail. HelmGuard says each finding can preserve the evidence examined at the time, the reasoning behind the conclusion, a confidence score, timestamps and any human decision or escalation that followed. Because the platform is built around a unified data model, the company says organizations can later determine what information was available, what the agent concluded and what people subsequently did without having to reconstruct the sequence from scattered records.

For auditors, HelmGuard's founders see that eventually shifting some work away from sampling documents and repeating manual checks toward examining verification records and the methodologies that produced them. They do not present the technology as a replacement for auditors. Their argument is that it could give auditors a more direct record for determining whether a control worked, what evidence supported that conclusion and how the organization responded.

When Assurance Crosses the Company Boundary

HelmGuard's larger wager is that the same machinery can eventually work between companies through what it calls the Verified Risk Network.

The model rests on separating a fact from the judgment made about it. Whether a vendor enforces multifactor authentication on its production systems, for example, can be verified as a fact. Whether that configuration is sufficient for a particular customer depends on that organization's risk appetite, regulatory obligations, criticality thresholds and use of the vendor.

HelmGuard is designing the network so underlying evidence can be verified and then mapped against whichever requirements an organization needs, including frameworks such as ISO 27001, SOC 2 and NIST and regulations including DORA and NIS2. Each organization would retain its own thresholds, criticality tiers and escalation rules. In other words, verification can be shared without requiring every organization to reach the same risk judgment.

Making that work between companies creates another problem: some of the strongest evidence of a control's effectiveness may also be information an organization has no intention of handing to its customers.

HelmGuard says it is developing zero-knowledge verification intended to allow a verified signal to cross organizational boundaries without the underlying configurations, logs or internal documents leaving the owner's environment. Counterparties would control what information is shared and with whom, while maintaining a verified profile that could be reused when other customers ask assurance questions.

The company is starting in environments where those trust boundaries are already relatively well defined, including conglomerates, private-equity funds and their portfolio companies, before extending the model outward. The founders acknowledge that the sequencing matters and that there is no immediate path from today's document-heavy assurance programs to continuous agent-to-agent verification across companies.

In the short to medium term, they expect agent-led assurance to supplement traditional GRC processes. Over a longer horizon, their expectation is that agents would perform much of the continuous collection and verification underneath the assurance function, while people would remain responsible for objectives, risk appetite, boundaries and consequential decisions.

Whether companies, auditors and regulators ultimately accept that arrangement at scale remains unresolved, particularly when automated verification reaches across organizational boundaries and into sensitive systems. HelmGuard is still building toward that model, and the Verified Risk Network it describes is considerably more ambitious than making today's assurance process faster.

That ambition is also what makes the company's $7.3 million seed round worth watching. HelmGuard is betting that AI will not merely help GRC teams complete the familiar work more efficiently. It is betting that continuous verification can change what organizations regard as assurance in the first place, shifting the center of gravity from what a company has documented about its controls toward what can actually be established from the evidence.

The GRC Report is your premier destination for the latest in governance, risk, and compliance news. As your reliable source for comprehensive coverage, we ensure you stay informed and ready to navigate the dynamic landscape of GRC. Beyond being a news source, the GRC Report represents a thriving community of professionals who, like you, are dedicated to GRC excellence. Explore our insightful articles and breaking news, and actively participate in the conversation to enhance your GRC journey.