Shift to exposure priority: The "post-Mythos" AI threat landscape requires moving from counting isolated CVEs to prioritizing connected exploit chains based on business impact and what to fix first.
AI-driven infrastructure: Brinqa uses specialized AI agents to deduplicate hundreds of millions of overlapping daily scanner findings and automatically infer asset ownership for the 70% of organizations lacking clean data.
Business-line calibration: Rather than using generic global risk scores, vulnerabilities are weighted by specific business-line P&L, revenue importance, and life-safety operational constraints.
Deliberate non-execution: The platform operates strictly as a decision-support graph, packaging remediation data for trusted deployment tools rather than maintaining direct write access to production environments.
In the middle of what’s already being called the post-Mythos stretch of the threat landscape, I had the chance to sit down with the leadership team from Brinqa at Black Hat to talk through where enterprise exposure management goes from here.
For background, the “post-Mythos stretch” of the threat landscape refers to the period after Anthropic’s Claude Mythos model demonstrated it could autonomously find and chain zero-day exploits, including one OpenBSD flaw that had gone unnoticed for 27 years of human review, at a speed no human red team, let alone blue team, had ever had to plan around.
That backdrop mattered to the whole conversation, because it's exactly the environment Brinqa has spent the better part of a decade building for, whether or not anyone had put a name to it yet.
CEO Dan Pagel framed the shift in language sharp enough to build a strategy around: CISOs aren't asking "what do we have" anymore, they're asking "what do we fix first, and how do we prove it mattered."
Those are two very different questions, and most of the vulnerability management category was built to answer the first one, not the second. When the volume and speed of exposures both go vertical at once (which is exactly what an autonomous, exploit-generating AI model does to a threat landscape), the industry's old answer, a scored list of everything that's wrong, stops being useful.
What a CISO needs instead is a ranked, defensible answer to what to fix right now, and a way to show the board that the fix actually reduced risk.
From Counting Vulnerabilities to Mapping Exploit Chains
Traditional vulnerability management treats every CVE as its own island, scored in isolation on a severity scale that has little to do with whether an adversary could actually reach it. Brinqa's pitch, and the thinking behind its integration with attack-path validation firm Horizon3.ai, is that risk lives in the chain, not the individual finding.
A "low" severity misconfiguration sitting two hops from a crown-jewel asset can matter far more than a "critical" CVE stranded on a box nobody can reach. That's the industry-wide shift from vulnerability management to what GigaOm and others now call exposure management: less about counting problems in isolation, more about mapping which combinations of them an attacker could actually walk through, start to finish.
That shift also explains why Brinqa avoids being just another scanner. The platform's job is to sit on top of the scanners an enterprise already runs, aggregate and enrich what they find, and resolve it all against business context, asset ownership, exploitability, and criticality, into a single, prioritized answer rather than another disconnected dashboard.
The Ownership Problem Nobody Wants to Admit
Ask most large enterprises who owns a given exposure and the honest answer, more often than not, is nobody knows for certain.
In Brinqa's own experience, somewhere around 70% of customers lack clear, reliable asset ownership data, which means the best prioritization engine in the world still stalls out at the last step: somebody has to actually be assigned the fix.
Brinqa's AI Attribution Agent exists specifically to close that gap, inferring asset and application relationships, business hierarchy, and accountable owners from the data the platform already has, at a confidence threshold in the neighborhood of 90% before it auto-accepts an attribution. It's an unglamorous problem to solve, but it's also the one that determines whether a beautifully ranked list of risks actually turns into remediation or just sits in a queue nobody feels responsible for.
When "More Data" Becomes the Problem
The other side of the AI coin is volume. Brinqa's own findings counts have surged from something on the order of 20 to 30 million a day to around 400 million over a recent four-month stretch. The platform's near-term infrastructure target is the capacity to handle on the order of three billion findings in a single environment without falling back to daily snapshots.
That kind of growth doesn't just strain compute, it strains meaning: when scanners, cloud posture tools, and AppSec pipelines all surface overlapping versions of the same underlying issue, the noise itself becomes a risk. Brinqa's AI Deduplication Agent, and the deterministic and semantic deduplication modes behind it, exist to consolidate that overlapping output into a single trustworthy record before a human or another AI agent ever has to act on it. It's the unsexy plumbing work that has to happen before any of the more interesting prioritization or automation conversations are even possible.
The Same Score Doesn't Mean the Same Thing to Every Business Line
Even after a finding is correctly attributed and deduplicated, a single global risk score can still mislead the people who most need to act on it. "High residual risk" reads the same on a dashboard whether it's attached to an internal wiki nobody outside IT has opened in years, or to the checkout service carrying the majority of a retailer's revenue.
Without business-line context layered on top, one pretty graph tells two different executives two different stories, and the one running a revenue-generating line of business has no way to tell, from the graph alone, whether the number in front of them is a rounding error or a threat to this quarter's numbers.
The stakes climb even higher in industries where a line of business isn't just generating revenue, it's keeping people alive. A hospital network, a utility running a power grid, a manufacturer operating industrial control systems—these industries don't have the luxury of treating a life-safety system, an infusion pump network, a grid control interface, a plant's safety instrumented system, as just another node on the graph.
The same generic severity label that might cost a retailer a bad quarter can, in one of these environments, get someone hurt if it's misjudged in either direction: too aggressive, and a critical life-safety system gets taken offline that never should have been touched; too passive, and a genuinely dangerous exposure sits unaddressed because it looked like every other routine finding in the queue.
That's the gap Brinqa's leadership flagged as one of the next six months' priorities: business-line risk modeling that weights the same underlying finding differently depending on which part of the business it touches, what that part of the business is worth, and what happens if it goes down.
Done well, that kind of customization does two things at once.
First, it lets the leader of an actual revenue-generating line of business make their own call on remediation priority and timing, because they're finally looking at risk expressed in terms of their own P&L, not a generic severity label borrowed from a security team that's never seen their revenue numbers.
Second, it gives that same leader something coherent to bring upstairs: instead of translating "critical," "high," and "medium" into a hallway argument about what it actually means, they can walk into a board conversation with a residual-risk figure that's already been calibrated to what's at stake in their part of the business, sitting on the same scale the board is using to weigh risk everywhere else it has oversight.
That shift is a relief as much as it is a capability.
Executive leaders and the owners of each line of business can stop guessing at what a generic severity label actually means for their part of the enterprise and finally exhale, because they're making genuinely business-informed decisions, backed by a risk figure calibrated to the dollars, customers, and regulatory exposure that are actually theirs to manage.
Skip that layer of customization, or work with a platform that isn't built to provide it, and most enterprises are stuck with the status quo: one flat, unweighted graph that can't distinguish between findings that matter and findings that don't.
A vulnerability sitting in the payment authorization pipeline and a near-identical one sitting in a mothballed internal expense-reporting tool can carry the exact same severity label on that graph, but they are nowhere close to the same risk, and a platform that can't tell them apart ends up alarming the teams that can least afford to be distracted while reassuring the ones that most need to be worried.
Push that miscalibration one step further into automated remediation and it stops being a communication problem and becomes an operational one. A security or SOC analyst working a queue sees an alert that looks like a quick, easy fix, not the revenue-generating business humming along behind it, and with no business context attached to the finding, that analyst can go ahead and take it down without ever knowing what they just did: who in their right mind would bring down a massive, revenue-generating line of business over what turns out to be a very low-risk alert?
That's a risk tolerance decision that belongs to the business leader who owns the P&L, and ultimately to the board that sets risk appetite for the enterprise, not to the analyst working the ticket queue.
The reverse holds just as true: give that same analyst accurate business context and impact data, and a genuinely high-risk finding sitting in a revenue-critical line of business gets escalated and acted on with the urgency it deserves, instead of getting lost in a queue behind a hundred other alerts that look identical without that context.
It also puts the CISO in the driver's seat rather than the back seat: instead of being the executive who shows up after the fact to explain why something got shut down unnecessarily, or why something serious got missed, they become the one who arms business leaders and the board with the calibrated risk picture that made the right call possible in the first place. This is a demonstrably more valuable role to play, and a bigger seat at the boardroom table to hold. And let’s not forget the high-fives that start coming the CISO's way from the CTO and CIO, who are invariably standing right next to them, for better or worse, whenever one of these calls plays out.
Why Brinqa Won't Pull the Trigger For You
One of the more counterintuitive things about Brinqa's roadmap is what it deliberately doesn't do: take direct remediation action inside a customer's environment. That's not a capability gap, it's a design choice, and it lines up with something close to a universal preference among the CISOs Brinqa works with. They want a platform that tells them exactly what to fix and how, but want the actual keystrokes to happen through their own tools.
Brinqa's answer is to package remediation recommendations for the systems that already have execution authority, tools like Automox and NinjaOne, rather than becoming another system with standing write access to production.
The same logic shows up in Brinqa's Bring Your Own AI model, which lets a customer plug in whatever AI agent or LLM they already trust, via API or Model Context Protocol, and have it operate against Brinqa's governed CyberRisk Graph rather than a black box. Given that roughly 80% of the patches enterprises actually need are third-party, and often require an interim mitigation rather than a clean patch, Brinqa decides what and why, the customer's own tools decide how. That division of labor is doing a lot of quiet risk management of its own.
The Real Gap Isn't the Platform, It's the Pitch
Here's the part of the conversation that should be the most useful to anyone selling technical differentiation into a skeptical enterprise: by its own account, Brinqa wins something close to nine out of ten proofs-of-concept it enters, and still isn't consistently named among the top three vendors CISOs list off the top of their head.
That's not a product problem, it's a translation problem. It's the gap between what a platform can prove hands-on and what a buyer can recall unprompted, and it's a gap that shows up long before a CISO ever picks up the phone to schedule a POC.
A GigaOm Radar placement helps close part of that gap (read a recent Radar report with Brinqa here) because it's third-party and comparative, but it doesn't replace the harder work of translating "we correlate roughly 270 connectors' worth of findings into a single risk-scored graph" into something a CFO or a board member can repeat back in a hallway. That translation, from technical proof to business-line language, is exactly the kind of work Brinqa and GigaOm are now scoping together, starting with CXO decision briefs and technical validation briefs built around the post-Mythos differentiation story.
Key Takeaway
The vulnerability management category spent two decades getting good at answering "what do we have." Brinqa's bet, backed up now by an independent Leader and Fast Mover placement in GigaOm's own Radar, is that the question that actually matters going forward is "what do we fix first, and how do we prove it mattered," and that answering it requires ownership attribution, deduplication, and exploit-chain context stitched into one graph rather than scattered across a dozen dashboards.
The platform is deliberately built to stop short of pulling the remediation trigger itself, leaving that authority with the customer's own tools, which turns out to be exactly what CISOs want from a vendor operating this deep inside their environment.
What Brinqa hasn't fully solved yet is narrative: a company that wins something like nine out of ten POCs but isn't reliably top-of-mind pre-POC has a translation problem, not a product problem, and closing that gap is the explicit work ahead for the next two quarters.
If you're a CISO, VP of vulnerability management, or anyone evaluating exposure management platforms in the post-Mythos threat landscape, this is worth watching closely. I'll follow up as Brinqa and GigaOm scope the joint research and validation work discussed in this conversation.
Reach out if you want to compare notes on exposure management in the meantime.
Rob Teel is a Field CTO and Senior Advisor for organizational technology strategy, innovation and digital infrastructure to drive business growth, competitiveness and transformational leadership. He is a Data, Analytics, AI, Security and Risk Expert in the Information Technology, Government, Defense, Energy, Telecommunications and Healthcare industries. Click here to connect with Rob or schedule an alignment strategy session.
Sources:
Executive Change: Brinqa Appoints Ron Dovich as Chief AI and Automation Officer, intelligence360
Anthropic's Mythos Has Landed: Here's What Comes Next for Cyber, Dark Reading