How one IT leader is budgeting for AI, security and vibe coding: Ready.Set.Midmarket! podcast

Demetrios Xarhoulakos explains how his lean team is approaching AI investment, governance and security as it plans the 2027 budget

"Vibe coding it with a mix of DevOps."

That's how Demetrios Xarhoulakos, director of IT at DeSimone Consulting Engineering, described one way his lean IT team is experimenting with AI to reduce costs and build capabilities internally.

But behind the catchy phrase is a much bigger challenge that IT leaders like Xarhoulakos are facing right now:

How do you budget for a technology changing faster than the budget cycle itself?

Xarhoulakos joined the latest episode our sister brand MES Computing's Ready.Set.Midmarket! podcast to talk with hosts Adam Dennison and Samara Lynn, along with Computing's editorial director Tom Allen, about DeSimone's 2027 technology planning and how AI is changing the budgeting equation.

AI spending meets IT budget

DeSimone, a global structural engineering firm with roughly 800 employees, operates with an IT team of about 10 people. Xarhoulakos said cybersecurity and AI are two major priorities as he begins mapping out the company's 2027 budget.

AI, however, presents a particularly difficult planning problem.

Heading into 2026, Xarhoulakos said his organisation had little dedicated AI spending beyond a small budget for an engineering-focused pilot. Actual adoption moved much faster than expected.

"I think that speaks to the scale at which this is moving," he said. "It's a challenge to plan for."

Now his team is looking not only at what AI costs, but where those investments belong in the budget and whether AI can offset spending elsewhere.

One example involves a contract renewal. Xarhoulakos said DeSimone is exploring whether some functionality the company currently pays for can instead be built internally, with employees collaborating and "vibe coding it with a mix of DevOps."

The ultimate goal, he explained, is to offset some AI investments by making other areas of the technology stack more efficient.

Security is still king

While AI is moving fast, Xarhoulakos said cybersecurity remains a bigger investment for DeSimone.

That mirrors findings from MES Computing's IT spending intent research, in which 72% of survey respondents identified cybersecurity and cyber resilience as top spending priorities.

For Xarhoulakos, the real challenge is demonstrating security ROI to business leaders when the investments are doing exactly what they are supposed to do.

"When things are working and...security's working well, it's kind of hard to notice that," he said.

As other IT leaders can probably attest, IT is the one department that comes under a fixed spotlight when any facet of operations goes sideways.

That scrutiny also means that existing vendors don't always remain part of the stack. Xarhoulakos said contracts are evaluated as they come up for renewal, including a phishing security platform due for review in early 2027.

Governance before AI scales

Xarhoulakos's advice to other IT leaders preparing their 2027 budgets centres on governance and the importance of keeping it in pace with AI adoption.

DeSimone has developed an acceptable-use policy and governance around issues including data sovereignty and AI consumption. HR also worked closely with IT on policy and governance.

"Governance is so important," Xarhoulakos said. "Getting people trained in your investments, getting governance in place, working on the security side of it, ensuring that your ducks are in a row is the biggest thing."

That discipline extends to AI costs. DeSimone is monitoring usage and tokens so IT can better estimate what growing adoption will cost.

And even then, Xarhoulakos said investment has to prove itself.

In his sector, engineering, that could mean asking whether technology can reduce a 40-hour task to 30 hours without sacrificing the reliability required when work involves something as consequential as designing a structural component of a building.

It also means watching whether employees continue using an AI tool after the novelty wears off.

"The ones we are interested in, the ones that we know we want to make that investment in, is the [AI tool] that people can't get away from," he said.

Watch the full episode of Ready.Set.Midmarket! now for the discussion on 2027 budgeting, cybersecurity, AI governance, training, ROI and how one organisation is trying to keep pace with rapid AI adoption

Watch the full episode on YouTube or listen on Spotify and Apple Podcasts.

This article originally appeared on MES Computing.