Dr. TJ Jiang, Chief Executive Officer and Cofounder of AvePoint.
getty
We’ve reached a pivotal moment: agentic AI tools are now advanced enough to operate with minimal oversight, propelling us into a new era of large-scale adoption. However, this independence is a double-edged sword. While it allows for more output with less human labor, it also introduces hard-to-catch vulnerabilities, gaffes, security breaches and other costly errors.
This is the trust gap: workflows are becoming more efficient, but their outputs are becoming less reliable. To bridge it, enterprises must move beyond deployment and build a dedicated “trust layer” into every agentic workflow.
The trust gap is a known problem. Gartner predicts that by 2028, 50% of organizations will adopt a zero‑trust posture for data governance as unverified AI‑generated data proliferates. Some companies have responded by introducing tools to monitor agentic output in an attempt to address this trust gap. Anthropic, for example, introduced a product that can check and proof AI‑generated code. Tools like this are helpful, but they're insufficient. They offer a fragmented view of risk. To secure agentic workflows, organizations need total visibility into agent behavior, activity and security across the entire enterprise, not isolated controls.
Instead of relying on a patchwork of point solutions to monitor and govern agentic AI, what organizations really need is a comprehensive trust layer. This framework extends proven governance, identity and access controls to every AI agent across all workflows and cloud environments.
This approach is critical as enterprises face increasingly complex issues like AI hallucinations, unauthorized data access and unmonitored agent activity that threaten both operational integrity and regulatory compliance. This is a critical necessity, as my company found that over 80% of organizations have had to delay AI deployments by up to a year to address data security and management risks.
A modern trust layer allows enterprises to:
• Monitor agent behavior at scale, tracking thousands of agents simultaneously to detect anomalous behavior before it escalates.
• Enforce granular policy, restricting sensitive data access to agents with validated credentials.
• Ensure compliance, maintaining an auditable trail of every AI-driven action to meet regulatory mandates.
There’s been a lot of chatter about AI killing SaaS or workflows being “agented away.” While it’s certainly true that agents will (and already have) reshaped SaaS and certain workflows in other industries, it’s also true that these agents need oversight and governance to succeed. The trust layer provides this necessary layer of governance, making the transformation of SaaS and other industries possible and sustainable at scale.
We should all be excited about the recent advancements in agentic AI, but we should also be honest: the very capabilities that make AI agents powerful—automation, autonomy and scale—also introduce new vulnerabilities, including hallucinations, governance gaps and increased risks to security and compliance.
This challenge demands more than point solutions; it calls for a comprehensive approach to governance and monitoring, ensuring every AI action is accountable and aligned with organizational values. In practice, many organizations are exploring more integrated approaches that can address data protection, AI governance and compliance in tandem, while also building the enablement and support structures that foster trust, security and productivity with AI.
With thoughtful leadership and a commitment to comprehensive oversight, enterprises can harness agentic AI confidently, building a future where trust is embedded at the heart of technological progress.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
此内容由惯性聚合(RSS阅读器)自动聚合整理，仅供阅读参考。 原文来自 — 版权归原作者所有。