Welcome to this week’s edition of CIOsurge!

This week:

Let’s make this week a game-changer.

Stay sharp. Stay ahead.

This week on the CIOsurge podcast I sat down with Gary Flowers, CIO at Year Up United, the national workforce development nonprofit with 35,000 alumni. Gary runs both technology and curriculum there, and he gave me the cleanest framework I have heard for thinking about AI inside an organization: split it into two buckets, because they need different expectations and different resourcing.

The first bucket is AI as a workforce skill. "Everyone in the organization should have some level of AI proficiency," Gary told me. This is the individual layer, and the returns are real but modest. "From a workforce skill standpoint, maybe it'll save me 10, 15% of my time. It'll make my presentations more accurate, more thought-provoking."

The second bucket is AI as an organizational capability, and this is where the transformation lives. "It may change departments," he said. It allows the organization to double down on the human advantage, remove the noise, and process information at a depth that produces things "we just would not have come up" with. His point: "Your expectations of both should be different. Your resources should be different, but you do need to kind of break it out into those two things."

Then he pushed the idea somewhere most leaders have not gone yet. Agents are workforce. He referenced a talk by Peter High that stuck with him: a 500-person company that would have grown to 700 people might stay at 500 and run 200 agents. "Well, who's managing that? That's still workforce. Those agents are workforce. Who manages that? Does the CHRO, the Chief People Officer, manage that? Does the CIO manage that? Someone has to manage it. Who spins up an agent? Who spins down an agent when it's not meeting metrics?"

Nobody has clean answers to those questions yet, and that is exactly why the two-bucket framing matters. Individual proficiency you can train. Organizational capability you have to architect, govern, and assign ownership for, including ownership of a digital workforce that does not show up in your HRIS.

And Gary's warning on the individual side is the one to repeat to every team: "AI is going to come take your job. No, what's probably going to happen, more realistically, is that the person that is using AI effectively is going to take your job."

The takeaway for technology leaders: budget and govern AI as two separate programs. Fund workforce skill broadly and cheaply, fund organizational capability selectively and seriously, and decide now who owns the agent headcount question, because it is coming to your org chart either way.

Book a meeting with me

Enterprises have poured spending into agentic AI in 2026, but readiness for full-scale deployment remains far off for most organizations, according to a Deloitte report covered by CIO Dive. Although roughly half of technology leaders say they have a clear view of their future AI operating models, only 15% of organizations have actually been able to scale multiagent systems. Most reported being unprepared across the areas that matter for real deployment: workforce, business processes, ecosystem partnerships, security, and governance policies. As China Widener of Deloitte framed it, the shift is not about adding more agents but about changing the way work happens, and it is a multiyear journey that will keep evolving as the underlying technology matures.

The gap between "we have a strategy" and "we have scaled" is the whole story here, and it is where most of this year's budgets are quietly stranded. Fifteen percent is a sobering number when you set it against how much has been spent. The organizations that close the gap will not be the ones that bought the most agents, they will be the ones that did the unglamorous readiness work first: mapping processes, fixing governance, and preparing the workforce. Gary's two-bucket framework in this edition is a good starting lens. If you cannot say who owns your agents and how they are measured, you are in the 85%.

— Zack Tembi

The bottleneck for scaling agentic AI is not the models, it is the data underneath them, according to a report covered by CIO Dive. "Without a deep understanding of the full data estate, including unstructured dark data, you do not have the quality and trust to activate agents at scale," the report's foreword said. The shift from AI that answers questions to AI that takes actions raises the stakes: agents need data from across the enterprise, structured and unstructured, carrying the right business context, and they need frictionless access to operational systems like supply chain, point-of-sale, and HR data to make decisions and act in real time. Limited access to that data, along with the necessary overhauls to business processes, workforce, and governance, is what stands between most organizations and agents that work at scale.

Every agent ambition eventually collides with the state of your data, and this is the collision. It is not glamorous, and it does not demo well, which is exactly why it keeps getting deferred in favor of the next pilot. But an agent taking action on incomplete or untrusted data is worse than no agent at all, because it acts confidently on the gap. Before you green-light another agent project, ask what data it will touch, whether that data is trustworthy, and whether the agent can actually reach the operational systems it needs. If the honest answer is no, the data layer is your real roadmap for the next year.

— Zack Tembi

Read the full story

Salesforce says activated agents nearly tripled in a year, with creation time down 53%.

Gartner projects Fortune 500 firms running 150,000+ agents by 2028, only 13% call governance adequate.

CIOs are being told to build executive AI literacy at the board level.

Regulators now demand agent audit trails, and most enterprises can't produce them.

Nearly half of orgs hit an AI data exposure incident, but only 34% have standards for how agents access company data.

40% of orgs reported inaccurate AI outputs in the past year, 27% a data breach, and 26% regulatory action tied to AI use.

Send us your story

CIO, American Financing

CIO, American Structurepoint

CIO, PDI Healthcare

vCIO (Healthcare), Meriplex

CIO, nVent