Priam Cyber AI

Priam Cyber AI is a London-based cybersecurity company that develops an AI-powered autonomous alert investigation platform for Security Operations Centers (SOCs). Its core product is AVA, an autonomous L1 triage agent designed to investigate security alerts and deliver defensible verdicts rather than simply summarizing or automating pre-written playbooks. AVA operates using a proprietary reasoning engine called PEBRE (Probabilistic Evidence-Based Reasoning Engine), which is designed to weigh evidence mathematically rather than relying solely on large language model (LLM) outputs. The system gathers evidence, tests hypotheses against the customer's environment, and produces one of three verdicts: true positive, false positive, or inconclusive — the latter explicitly naming what evidence is still missing. Each alert processed by AVA results in an Investigation Report containing the verdict, the evidence chain, the reasoning connecting them, and any identified gaps. The platform targets enterprise SOC teams and MSSP/MDR providers operating across cloud, on-premises, or private environments. AVA integrates natively with common security stack components including: - Microsoft Defender for Endpoint and Microsoft Sentinel - CrowdStrike Falcon - Palo Alto Cortex XDR - IBM QRadar - Elastic Security Observed deployment metrics cited by the company include 97% false-positive closure with defensible reports, reduction of L1 triage time from 35-60 minutes to 1-2 minutes, and L2 investigation time from 150 minutes to 24 minutes. The platform is positioned as a distinct category from detection (SIEM/XDR) and automation (SOAR), focusing specifically on the investigation and conclusion phase of alert handling.