AI Observability Platform Evaluation Criteria for Enterprises
Legacy monitoring can't track whether AI agents called the wrong tool or skipped required steps.
Tariq Osei-Mensah
Senior Editor, AI Infrastructure
Tariq Osei-Mensah
Tariq covers the architecture and governance layer of autonomous AI systems, drawing on a decade of prior work as a distributed-systems engineer at enterprise software firms. He joined the AI security beat in 2019 and has since become a go-to voice on control plane design and agent orchestration risks.
11 stories
Detecting Compromised or Hijacked AI Agent Sessions
Hijacked agents succeed silently, completing the wrong task without raising alarms.
Tariq Osei-Mensah
Integrating AI Agents With Microsoft Entra ID
Tariq Osei-Mensah
Integrating AI Agent Authentication with Okta
Okta treats AI agents as distinct identities with enforced credentials and continuous oversight.
Tariq Osei-Mensah
Shadow AI Examples Across Enterprise Departments
Source code and financial data leak fastest into unsanctioned AI tools across departments.
Tariq Osei-Mensah
SDK generation tools for AI gateway and control plane APIs
SDK generators for AI control planes must prioritize OpenAPI fidelity and air-gapped output.
Tariq Osei-Mensah
Policy Enforcement at the AI Control Plane Layer
Enforce AI guardrails at the network layer, not scattered across applications.
Tariq Osei-Mensah
Real-Time Observability Telemetry in an AI Control Plane
Observability becomes the operational foundation for governing autonomous agents at scale.
Tariq Osei-Mensah
enterprise AI control plane platforms compared
Platforms differ wildly in how they govern agent access and detect threats.
Tariq Osei-Mensah
AI Control Plane Architecture for Enterprise Deployments
Enterprises must build unified control planes to govern AI spending and prevent shadow deployments.
Tariq Osei-Mensah
Shadow AI Detection in Enterprise Networks
Organizations must layer network, browser, identity, and behavioral detection to catch shadow AI.
Tariq Osei-Mensah