It would be amazing to have this kind of capability (albeit smaller scale) for personal use on home networks and labs. All inference running locally (as an option)
We've been working hard over the past few months on full autonomous closing/escalations of high confidence items via agent swarms.View quoted post on X
Right now it's in monitor mode, but because our model is working so effectively, the success rates are staggering.
If not in monitor mode, NightBeacon would have automatically closed 127,928 alarms and escalated 978 - all within a sub 30 second period of time. Meaning, things you need to respond to are already there for you versus sifting through multiple alarms.
We also bulk alarms into "Situations" - meaning 500 alarms can be one situation with all the context, root cause analysis, agentic discovery of evidence, and verdict all right there for an analyst.
We're monitoring the data for a few weeks/month to ensure there isn't any glitches in the system, but its extremely promising.
Thats a true signal to noise ratio. Affording time back to the SOC to be more proactive.
Right now, talking about an overall 87% reduction in volume for analysts, and we're only getting that number better.
#NightBeacon #BinaryDefense