AI is forcing IT deeper into the content stack because AI runs on information, not aspiration.— Chris McNulty
I have been thinking a lot about AI and information over the past two weeks. What I saw in Orlando and Baltimore pointed to the same conclusion from two very different angles.
At the Microsoft 365 Conference in Orlando on April 21 to 23, Microsoft and its partners laid out a future where AI is built into the flow of work. The conversation has already moved past basic generation. It is now about agents, custom skills, SharePoint as a system of knowledge, and the management layer required to keep all of it under control.
One week later at the AIIM Summit in Baltimore on April 28 to 30, I saw three distinct audiences in the room. That split tells us a lot about where information management sits in the AI transition.
First, there were the forward-looking teams. They want to understand what is possible with agentic document processing, and vendors like Hyland, OpenText, IBM, and Box were ready to show them.
Second, there were traditional information governance and records management practitioners who still seem to hope this wave will pass. It will not.
Third, there were information managers caught in the middle. They understand the risk of standing still, but many still do not have a clear operating model for moving forward.
01 — A Shifting Center of Gravity
A Shifting Center of Gravity
Information managers have traditionally sat closer to governance than to core IT. AI is changing that. The center of gravity is shifting, and fast.
AI is forcing IT deeper into the content stack because AI runs on information, not aspiration. That makes content readiness and content-process observability strategic requirements, not cleanup projects.
This is not a contest between information technology and information management. The winners will be leaders who can operate across both disciplines and connect governance, systems, and business outcomes.
That matters because the hardest questions will land in IT whether IT is ready or not.
The Questions That Land in IT
Why is the metadata wrong? Why did the agent cite the wrong contract date? Why can nobody explain which document is current?
Traditional data governance models do not answer those questions on their own, and most IT teams are not set up to manage the content layer in that level of detail.
02 — The Cost of Waiting
The Cost of Waiting
AI is opening real competitive distance between organizations that move and organizations that wait. If your plan is to spend the next 18 months cleaning data, documenting use cases, drafting specifications, and running cautious pilots, you may discover that the market did not wait for your governance program to finish.
The better model is to put AI specialists close to the business teams they support and learn in production, with guardrails. Recent research across the industry points in the same direction. Execution speed matters, but so does feedback from the people doing the work.
AI is still uneven. Some days it rewards speed. Other days it exposes every weak process behind the scenes. Precision is improving, but it is not evenly distributed.
03 — Aligned to the Business
Aligned to the Business
The way through this is simple to describe and hard to execute. Stay aligned to the business and follow the work that drives revenue, service quality, and risk reduction.
We have seen this pattern before with the internet, mobile, and cloud. Each shift began as a specialist conversation and ended as part of how every serious business operates. The organizations that resisted did not preserve the past. They gave ground to competitors.
That is why this is ultimately a business question, not a doctrinal one. Governance matters. Controls matter. But if you do not understand how the business makes money this year and how it plans to make more next year, your governance model is disconnected from reality.
That leaves traditional information managers in a stronger position than many realize. If they can connect business context, content quality, and technology execution, they can move to the front of the pack. The path forward is not to retreat from AI. It is to meet it with better operating discipline.
