Analysts got 19 minutes back every hour in Stellar Cyber’s agentic auto triage trials
An analyst opening a queue on Monday morning will spend most of it on tickets that amount to nothing. Stellar Cyber’s Agentic Auto Triage closed 8,047 of those tickets on its own during customer trials, filing them as confident false positives.
That covers 64% of every verdict the software issued. Another 1,875 alerts went up the chain as real threats. The remainder got logged as informational and left there.
The trials ran 124 days inside customer environments. Stellar Cyber evaluated 138,475 alerts in that window and traces each published figure back to end-of-trial reports the customers received.
19 minutes an hour, and what fills them
Stellar Cyber measured 19 minutes returned per hour per analyst. Over a week that comes to close to a day.
Lisa Liu, communications and corporate marketing manager at Stellar Cyber, said junior analysts “will be able to operate at the level of senior analysts.” She named the hiring squeeze as “the current bottleneck that the entire industry is feeling with the difficulty of finding appropriate staff to handle both alert volume and deal with the level of threats effectively,” and put the returned hours against it. The job that comes out the other side is one of “oversight and proactivity.”
Every vendor is selling autonomous AI
Stellar Cyber’s marketing tells buyers to be skeptical of autonomous AI claims, its own included. The company offers trial counts as the answer.
Liu said: “The industry is being flooded with noise about AI, and each claim sounds more extreme than the next. But the only way to differentiate between the tools that will prove to actually improve your workflows and supplement your analysts, instead of creating more work for them, are the ones that act symbiotically with your analysts and boost their productivity, instead of unsupervised autonomous agents that could cause more damage than they prevent. The burden of proof of autonomous AI claims should include logic behind the approach to show how the technology works in concert with human analysts, and not an unsubstantiated figure about improved efficiency.”
The agreement number
Humans and the agentic bots landed on the same verdict 99.7% of the time. A review step stays in the workflow anyway.
Liu called that step “the playbook for the foreseeable future,” citing “the sheer speed of AI-driven decision-making.” One missed call, she said, “could compound into a disastrous incident.” Her argument for keeping the step runs on caution: without it, assets “become liabilities.”
The noise starts upstream
The detection layer feeding the queue generated the alerts that triage then throws away. Triage absorbs that volume at machine speed.
Liu expects the volume to climb “as enemy agents also develop and acquire more and more sophisticated AI-driven tools to ramp up attacks.” Persistent alert noise, in her words, has always been “a particular pain point for the industry.” Handling the changed volume “will make the difference between success and failure for your business.”
How far up the stack machines get to act
Stellar Cyber’s Multi-Layer AI architecture stacks data transformation at the bottom, detection management above it, and remediation on top. Liu named the principle that governs how high the automation reaches:
“The thesis of our technology is full visibility; even at the most automated level, the human analyst still has agency and the ability to intervene or question the machine-made choices. We believe in giving both a breadth of view and a depth of explanation when it comes to automated decision-making. This visibility is what allows human analysts to step up their own capacities and eventually begin to wield their AI tools more effectively. The guiding principle has always been human-augmentation, both in the sense that the human is guiding and overseeing machine-led decisions, and also that the machine is empowering the human to act faster and smarter in every situation.”
If you’re at Black Hat USA 2026, visit Stellar Cyber at booth #5542 to learn more!