Cogent Safety Makes the Fortune Cyber 60 for a Second 12 months, With a Plan to Run Cyber Protection Like a Manufacturing facility

San Francisco, CA, November third, 2026, CyberNewswire

Cogent Safety, an utilized AI cybersecurity firm, reported that the PHP2Shell vulnerability uncovered important gaps in enterprise vulnerability response. Disclosed on October 21, PHP2Shell concerned two important flaws that had remained in PHP for greater than a decade. AI-assisted analysis helped determine the vulnerabilities and develop working exploits in roughly one week, with assaults starting earlier than a patch was obtainable. Cloudflare recorded tens of millions of exploitation makes an attempt that day.

In response to Cogent Safety’s analysis, by 5 p.m. Pacific, the typical Fortune 500 firm had patched fewer than 1% of its 2,266 affected belongings.

That hole between how briskly attackers transfer and how briskly massive corporations reply is the issue Cogent was constructed to shut. The corporate, an utilized AI lab targeted on cybersecurity, has been named to the Fortune Cyber 60 for the second consecutive yr.

The record, developed by Fortune with Lightspeed Enterprise Companions and AWS, acknowledges 60 of an important venture-backed cybersecurity startups which have but to go public, be acquired, or attain one other important exit, grouped by stage. Choice weighs monetary efficiency, progress, enterprise operations, and cybersecurity market analysis, and Fortune validates the ultimate record.

“Attackers already run like a manufacturing facility. AI is discovering the bugs and writing the exploits, and in some instances it’s working a lot of the assault,” stated Vineet Edupuganti, Cogent’s co-founder and CEO. “Most enterprises are nonetheless defending by hand, one ticket and one handoff at a time. Making this record two years in a row tells us safety leaders are prepared to vary how that work will get executed.”

The attackers constructed their factories first

Edupuganti borrows an analogy from manufacturing. Factories changed the lone expert employee with a manufacturing line, and essentially the most superior crops now run largely on their very own whereas individuals design the road and deal with the exceptions. Offense, he argues, has already made that shift.

The information factors the identical method. Google’s M-Developments 2026 report estimates the imply time to take advantage of at adverse seven days, which suggests exploitation now routinely begins earlier than a repair exists. Researchers have proven AI producing working exploits for newly revealed vulnerabilities in about quarter-hour, at roughly a greenback apiece. In a state-sponsored marketing campaign that Anthropic disrupted, AI brokers carried out 80 to 90% of the tactical work.

Enterprise protection nonetheless strikes on the tempo of its slowest handoff. When PHP2Shell landed, safety groups needed to discover each asset working PHP, together with base pictures and forgotten servers, then work out which have been internet-facing and which the enterprise trusted. After that got here homeowners, tickets, change home windows, and a rescan to substantiate every repair held, with each step falling to a talented particular person with a software.

Constructing the Defender’s Manufacturing Line

Cogent’s reply is what it calls the Cyber Protection Manufacturing facility, an working mannequin by which AI brokers take over the work that now strikes by hand between safety instruments, whereas individuals set the insurance policies these brokers observe. The corporate sells an enterprise platform for autonomous cyber protection, and its first product applies the mannequin to vulnerability administration.

The road begins with a list. Brokers construct one deduplicated report of each asset throughout cloud, on-premises and endpoint environments, inferring each’s proprietor and enterprise significance from methods the corporate already retains. Findings move in from the scanners and safety instruments clients already run, and Cogent matches newly revealed vulnerabilities towards its reside software program stock so publicity surfaces earlier than scanners catch up. Brokers then verify whether or not the weak software program is working and reachable, they usually write out the reasoning behind each precedence.

Cogent teams remediation work by repair, so one patch can resolve a vulnerability throughout dozens of servers. Contained in the autonomy zones every buyer defines, brokers perform supported fixes on their very own. On managed endpoints, as an illustration, they write the remediation script, push it by the corporate’s current endpoint administration platform, and skim again the outcomes. Modifications exterior these zones go to a single proprietor as a ticket and look ahead to an individual’s approval, and Cogent closes the work solely after the following scan confirms the repair held.

The corporate says the objective is for enterprises to remodel their cybersecurity features with a manufacturing facility mannequin to do 100 instances extra work, 100 instances sooner. Most clients begin with brokers in an assistive position and widen their authority because the outcomes earn belief, an strategy Cogent calls progressive autonomy.

PHP2Shell gave the mannequin a reside check. Cogent’s Zero Day Response processed the advisory and mapped weak belongings throughout buyer environments in a median of 5 minutes, days earlier than scanner signatures have been obtainable. Utilized to the typical Fortune 500 surroundings in Cogent’s analysis, the evaluation reduce 2,266 affected belongings right down to 16 that had PHP loaded at runtime, confronted the web, and supported important enterprise features. That shortlist was ready for safety groups after they sat down to start out work.

A Lab With Its Personal Fashions

Cogent describes itself as an utilized AI lab that simply occurs to concentrate on defensive cybersecurity, and the label displays how the corporate is constructed. It’s the solely cybersecurity firm that trains and deploys its personal frontier reasoning fashions for safety use instances. Its domain-specific mannequin, VR-1, has topped cybersecurity benchmarks, outperforming general-purpose frontier fashions on safety duties.

The mannequin works alongside Cogent’s Context Engine, which ties every vulnerability to the belongings, identities, functions, homeowners, and dependencies round it. That context normally sits scattered throughout asset databases, ticket histories and structure paperwork, and pulling it collectively is what lets brokers decide which modifications are secure to make.

“A general-purpose AI mannequin is aware of somewhat about all the pieces,” Edupuganti stated. “When an AI agent is deciding whether or not a patch is secure to push to some thousand manufacturing servers, it wants a mannequin that has discovered how vulnerabilities, exploits and enterprise methods behave, plus deep context on the corporate it’s defending. Put these collectively and you’ll give brokers actual accountability.”

Racing the Exploit Clock

For safety leaders, Edupuganti argues, the quantity that issues now’s the time between a vulnerability’s disclosure and a verified repair, and that window needs to be shorter than the time an attacker must construct and launch an exploit. Cogent is betting that closing it takes a distinct working mannequin, one the place safety work grows with the manufacturing line reasonably than with headcount.

The corporate’s second straight look on the Cyber 60 suggests consumers are warming to that guess, although Edupuganti expects individuals to remain on the middle of it.

“Probably the most superior factories nonetheless have individuals in them. They resolve what the road builds and the usual it holds,” he stated. “Safety groups are transferring into that position. They set the coverage and make the judgment calls, and the brokers do the work at machine pace.”

About Cogent

Cogent Safety is an utilized AI lab constructing the enterprise platform for autonomous cyber protection. Its AI brokers discover and repair vulnerabilities at machine pace so each enterprise can run its personal Cyber Protection Manufacturing facility. Beginning with autonomous vulnerability administration, Cogent carries vulnerabilities from disclosure to a verified repair throughout cloud, on-premises, and endpoint environments, figuring out publicity inside minutes and confirming what attackers can attain alongside the best way. Clients set the insurance policies the brokers observe and resolve which fixes run autonomously and which look ahead to approval. Fortune 500 safety groups utilizing Cogent have diminished the publicity window for important vulnerabilities by 97 %. Constructed by researchers and operators from Google DeepMind, Irregular Safety, and Coinbase, Cogent is backed by Greylock Companions and Bain Capital Ventures. Study extra at cogent.com.