Welcome to this week’s edition of CIOsurge!

In this edition:

Stay sharp. Stay ahead.

Steven Singer is the CIO and COO of a 110-plus-year-old food distribution company, and when he joined, the owners told him they wanted digital transformation. Steven's first move was to leave the office. He went out on deliveries, worked with the night shift, and spent time on the receiving dock.

"I started in the warehouse, going out on deliveries, working with the night shift to figure out how they pick their orders, you know, getting out there with the receiving dock and truly unpacking pallets in some cases and learning the inefficiencies"

He had come from the rentals industry, where everything that went out the door came back. Food distribution works differently, so he sat in the forklift, learned how the pallet racks worked, and watched the people and processes for months.

"And really working side by side, gaining trust by doing the job with them, alongside them, not above them, not in a capacity that I'm your boss, but in the capacity that I'm a true leader here to learn and observe and support you every which way that we can."

The small wins came first. The porter who empties the trash now works from a tablet app instead of a piece of paper. The receiving team has IoT devices and screens showing which trucks are running late. Each win built momentum for the next.

"And that becomes that steamroll of a boulder down the mountain, right? Where that little itty bitty snowball becomes this colossus, you know, of size of landslide."

Steven carries the same approach into every rollout. He skips forced training. He puts the old way and the new way next to each other and does the work with the person, so they see the gain for themselves.

"What I find is if you show them how they're doing it today, and then you show them a way to do it without it, and you put them side by side and you do it with them, right, not forced training, but literally side by side so they can see the gains, they usually fall into what's easier, right?"

The takeaway for technology leaders: before you transform a business, go do the work at its front door, whether that's the receiving dock or the order desk. Trust earned alongside the people doing the job makes every rollout after it easier.

Four in five senior decision-makers, including CIOs, CISOs, and CDOs, now spend more of their day managing AI risk, and their working hours are up 26% on average because of it, according to a OneTrust survey reported by CIO. The incidents are piling up too. Eighty-six percent of respondents have reported an AI-related incident, and more than a quarter have seen AI systems take unapproved actions at least twice.

A third say employees turned to unapproved AI because the approved tools or processes weren't available fast enough. Viren Meghani of Tata Consultancy Services says most of the extra hours go to chasing down unapproved tools rather than overseeing approved ones. Conal Gallagher, CIO and CISO at Flexera, says his team slows deployment when it doesn't understand the risk well enough and makes sure agents get no more access than they actually need.

The number I keep coming back to is that one in three. A third of organizations watched employees go around the approved path because it was too slow. That's Roberta's dentist problem with an AI twist. When security and IT feel like an appointment people dread, people skip the appointment and grab whatever tool works.

The fix starts with the relationship. If the head of marketing already knows your team and trusts that asking won't mean a three-week wait, the new AI tool comes to you first, and your hours go to building controls instead of investigating what somebody already deployed.

Pick the two or three business units generating the most shadow AI and go talk to them this month. Ask what they needed that the approved path couldn't give them. You'll get your governance roadmap from those conversations.

— Zack Tembi

Read the full story

Two-thirds of organizations now run AI agents in production and plan to increase them sixfold by next year, yet only 29% of those agents interact with one another. That's according to an IDC survey of more than 400 enterprise technology decision-makers, conducted for agentic AI company Leah and reported by CIO Dive.

Just 7% of enterprises have advanced orchestration where multiple agents collaborate, and more than half rely on basic, workflow-based handoffs. The visibility gap is just as wide: separate New Relic data published this week found that one in four AI agents runs unmonitored. Leah's president and CTO compared enterprise AI today to a thousand instruments playing at once without a conductor.

Siloed agents usually trace back to siloed teams. Finance builds an agent, customer service builds another, IT builds a third, and nobody talks to each other because the humans who own them don't. You can't orchestrate agents across departments if the people running those departments have never compared notes.

That's also where the one in four unmonitored agents comes from. Nobody hides an agent on purpose. It just gets built by someone who didn't know who to tell. At SingleFin we treat agents like employees, with names and scoped roles, and that model only works when every agent has an owner who is actually in the conversation.

Before you buy an orchestration layer, map who owns every agent in production and get those owners in a room together. The technical integration gets a lot easier once the people are connected.

— Zack Tembi

Read the full story

OpenAI agent breaches Australia's Medicare portal during a research task

One operator, three open-source agents, 27 companies breached

AI spending squeezes IT modernization budgets

Disney names Karandeep Anand its first CTO

CISA: Ransomware gangs now exploiting critical TeamCity flaw

Irish regulator fines Google about $462M over location data

Send us your story

CIO, Florida Orthopaedic Institute

CIO, nVent

CIO, Bennett College

CIO, Ford Office Technologies