Nigerian Founder Launches ADT, a New AI Model for Cyber Defense

Glemad, an AI security and research company, today announced Autonomous Defence Transformers (ADT) – a new class of AI models pretrained specifically for security reasoning and autonomous defense. This marks the first time frontier-scale AI models have been explicitly optimized for defensive reasoning rather than adapted from general-purpose systems.

“Security has always been reactive. We built ADT to change that – to give organizations the ability to reason about threats continuously and act at machine speed, without sacrificing safety or accountability.”

– David Idris, CEO and Founder, Glemad

Modern attackers operate at machine speed, automate reconnaissance, and adapt behavior dynamically. Yet most security systems remain fundamentally reactive – collecting telemetry after actions occur, triggering detections after damage begins, and gating response through human triage operating under time pressure. The result is a structural mismatch: attackers complete their objectives in the gap between detection and response, while defenders struggle to maintain security invariants over infrastructure state.

Since 2020, Glemad has pursued a fundamental thesis: security should be treated as a continuous reasoning-and-control problem, not an alerting-and-documentation problem. This led to ADT – AI systems structurally different from conventional security approaches. ADT models are trained on security-specific corpora from the ground up. Attack telemetry, compliance frameworks, infrastructure semantics, and adversarial tradecraft are embedded as first-class concepts during pretraining – not added later through fine-tuning.

PulseADT, the operational implementation of ADT models, is currently protecting over 680,000 assets while processing 1.8 million security events per second.

• Mean Time to Detect (MTTD):0.8 minutes vs. 287 minutes industry average (359x faster)

• Mean Time to Contain (MTTC):3 minutes vs. 420 minutes industry average (140x faster) These results demonstrate ADT’s ability to reason continuously over infrastructure state, interpret threats under uncertainty, and execute validated containment faster than humanpaced response allows.

The latest release introduces reinforced capabilities across three dimensions:

• Signal Interpretation: Processing operational text and structured telemetry to identify key events within high-volume logs

• Deviation Detection: Identifying changes in behavior across time and systems, including subtle distribution shifts

• Dependency Reasoning: Mapping identity chains, permission structures, and crossservice dependencies.

Alongside PulseADT, Glemad launched the Glemad API – giving developers programmatic access to ADT capabilities across five functional areas:

• Threat Detection API: Real-time classification using ADT Neural, ADT Anomaly, and ADT-4 Pro engines

• Vulnerability Scan API: Full ADT assessment pipeline across network, web, and TLS layers

• Threat Intelligence API: Continuously updated feed of IOCs, malicious IPs, and domain reputation

• Compliance Report API: Audit-ready reporting against GDPR, HIPAA, and PCI-DSS

• Active Defence API: Autonomous containment within policy constraints.

PulseADT is available now at getpulseadt.com with a 15-day free trial. The Glemad API is accessible via console.glemad.com with tiered pricing starting at $29/month for developers.

Glemad is an AI security and research company building auditable AI systems that understand, reason, and act on threats – reliably and autonomously. The company is committed to long-term research spanning decades: better reasoning under uncertainty, more robust consensus mechanisms, and stronger guarantees for autonomous action.