- Two Llama models run on the endpoint, alongside multiple detection engines for layered protection. - Train the AI on your own environment. The built-in LoRA training page makes that easy. - Add your own Sigma, YARA and signature rules next to AegisAI's to catch what generic rules miss. No more one-size-fits-all. - A plain-English storyline for every incident, fully air-gapped. - Fully customizable, built on open standards.