SyncAI.news, a Varaisys broadcasting
Why AI Agent Governance Must Be Built Into Enterprise Architecture
TB

Tim Bajarin, Contributor

· 1 min read

World NewsForbes: Innovation

Why AI Agent Governance Must Be Built Into Enterprise Architecture

As AI agents spread across the enterprise, IT directors I talk to are asking more questions about how to govern them.

With an incoming Democratic-controlled Congress, AI safety and accountability are also likely to draw greater scrutiny in Washington. But enterprises should not wait for Washington to determine how agents operate inside their organizations.

Over the years I’ve covered the technology industry, one pattern has remained consistent whenever a new computing paradigm emerges: Technology comes first, excitement follows, and controls arrive later—usually after something goes wrong.

That pattern played out with PCs, the internet, mobile and, more recently, the cloud. Now AI agents are on deck. But this time, I believe the gap between adoption and governance will need to close much faster. And it has to.

When Software Starts Acting On Your Behalf

The key difference between AI agents and tools used primarily to generate answers is that agents take action. An agent can open a support ticket, process a refund request, query customer databases, send emails and initiate workflows in other systems. Once software begins making decisions and taking action on a company’s behalf, it is no longer simply a productivity application. It becomes a participant in the organization—one that needs rules, limits and accountability.

In my conversations with CIOs and other technology leaders, a consistent theme is the use of AI agents across customer support, IT operations, finance and sales. Many of these proofs of concept have been developed by individual business units using different platforms.

This feels familiar. We saw a similar pattern with shadow IT as business units adopted cloud and SaaS services outside central IT oversight. The difference is that these shadow systems can take action.

Start With Identity, Ownership And Access

Consider what a properly run enterprise should know about every agent operating in its environment:

Original source

This story was published by Forbes: Innovation and written by Tim Bajarin, Contributor. SyncAI.news shows a preview; the complete article is on the publisher's site.

Read the full story on forbes.com

Similar News