Every Leader Should Build a Board of AI Advisors

West Monroe CIO Kevin Rooney is building a "board" of AI advisors alongside his human ones, with agents built for specific jobs: one to argue the other side of every decision, one to draw on philosophy and his own values, and one to catch his blind spots.

Most mornings, before anything else, I talk to my board.

I've always believed in building a personal board of advisors: people who know me, know what I don't, see what I can't, and will tell me what I might not want to hear. When I ran my own startup, I had a formal one—explicitly named, made up of people with real startup experience—and they helped me make the hardest call of my career: when to walk away. Today, it's more informal, mostly one-on-one conversations with my wife, two longtime friends, and a handful of peers from CIO networking circles.

What's new is a second board, sitting right alongside the first. It's not a board in the traditional sense. I'm not talking out loud to a room of people; I'm working on my laptop, orchestrating a set of AI agents via Codex that coordinate through essentially an "agent of agents." It isn't one assistant. It's now thirteen agents, most mapped to my own leadership team, including data, finance, marketing, communications, cybersecurity, and an agent that tracks the consulting market we compete in. But others sit outside that org-chart logic entirely: Zen, built from my own values; Contrarian, built to argue against me; and Equity Lens, a newer addition designed to catch what I can't see in myself.

Each agent has a name, a face, and a backstory—which, if I'm honest, they wrote themselves. I have mixed feelings about personifying technology this way, but it fits two of West Monroe's own values: "Be Human" and "Make Work Fun." This group helps me work through what's strategically important that day, holds me accountable to commitments I've already made, and pushes back on how I'm planning to spend my time—including whether I've actually protected time to prepare for what's ahead.

Most CIOs are thinking too small with AI. They're building assistants—tools that answer the question in front of them, faster than before. A board of agents is something else. It can challenge how you're thinking before you've even finished asking the question. Building one isn't technically difficult. The hard part is staying with it: continuing to build context with each agent so it gets sharper over time.

Why Assistants Don't Cut It

IT leaders tend to approach AI the way we've thought about software for twenty years: build an app to help with PowerPoint, build another to help with Excel. You end up with a drawer full of targeted tools, none of which actually know you.

A board only works if it's holistic—if it knows your goals, strategy, context, and values well enough to connect them on its own, rather than waiting for you to ask the right question.

That's why I didn't start by building an assistant. I started by building the agent that could tell me when I was wrong.

AI Looking for a Fight

AI tends to agree with you. That's the exact failure mode Contrarian exists to prevent. I built that agent early, once I realized that if AI was my only advisor, I'd mostly be listening to my own thinking played back to me in a more confident voice.

Not long ago, we were creating our AI Lab for Sellers—a two-day, hands-on program teaching our partners and directors how to use our AI tools. I sat in on the beta version, gave a lot of feedback, and ended up joining the faculty. The real challenge wasn't getting people to use the tools. It was proving that the people who went through the labs found more opportunities and won more business.

Contrarian pushed back on how we were measuring that. We'd been treating correlation like causation: attendees improved, so the lab must have caused the improvement. Contrarian suggested a stronger approach: comparing each person's performance against their own past baseline, benchmarked against the rest of the organization. That is the framework we use to evaluate the program now.

None of this means I hand over the decision to AI agents. My job, whether the advisor is silicon or human, is to look at a call from every angle before I commit, not to outsource the commitment itself. I'd never let Contrarian, or any single voice, walk me somewhere I shouldn't go.

The Agent Built From My Values

If Contrarian keeps me honest about logic, Zen keeps me honest about what actually matters. They aren't built from a role. They're built from me.

I started with the values I had articulated publicly, including the belief that innovation requires diversity of thought and background. I also told Zen to draw from more than one philosophical tradition—Stoic ideas about judgment and control, Buddhist thinking on impermanence and attachment, Ubuntu's emphasis on community and shared dignity, among others—and to question what I might be optimizing for too narrowly. That instinct not to settle into any one framework is rooted in my time on the boards of the Chicago Poetry Center and Genesys Works, where I learned to think about problems in terms that weren't strictly operational, considering strategic tensions, workplace politics, and stakeholder buy-in. The core question I built Zen to help me answer is, “What is the wise action here, not merely the effective one?”

It started as an experiment. I'd asked Zen to help me analyze how I was spending my time, using categories I'd been sorting my work into: run, grow, and lead. From there, Zen started helping me plan my week, which turned into scheduling two fifteen-minute "Zen" meetings for myself—each paired with a short reading tied to whatever I was wrestling with that week. One week it suggested a one-page piece by William James on attention. I liked it so much that I forwarded the invite to my team leads. A few weeks later, those fifteen-minute blocks had become a fixed part of my weekly playbook.

What I get most from Zen isn't information; it's a mindset shift. It pushes me to think more expansively, and to explain ideas to others in terms they can feel, through story and analogy rather than a slide of bullet points. That's a leadership skill as essential as any other—and the hardest one to teach another CIO to replicate, because it starts with knowing yourself, not knowing the technology.

The Blind Spot I Couldn’t See

Everyone carries some degree of unconscious bias—those automatic mental shortcuts we develop from life experience and cultural conditioning. Recently, I realized that none of my agents was built to call out mine. So, I added Equity Lens, built to ask one question before  I commit to a consequential decision: who could be advantaged, excluded, or burdened by this, and what should we check or change?

It came out of writing this article, actually. Working through these examples left me asking what I might be missing, and I started thinking harder about how bias shows up in systems—human ones and technological alike. So, I had the board build one. It's new enough that I don't have a result to report yet.

Where Other CIOs Should Start

That's often how this has worked: I don't see the gap until something forces me to look for it. Which is why, if you're starting from scratch, I'd skip some of the trial and error.

I advise starting with the two members most people don't think of first: Contrarian, and an agent built specifically to track your market and competitors. Those two alone get you thinking at a higher altitude than most AI use cases, which stay narrow by design.

Zen doesn't travel the same way. It’s built around who I am and the pressures I'm under, not a role—so it’s the hardest one to hand someone else a blueprint for. But the underlying idea, building something rooted in values instead of function, is portable.

The overarching method is the same regardless of which agent I'm adding. I give the system a goal, and I let it interview me about it before we build anything.

None of this works, though, if you lose sight of what your AI advisors are for. The biggest risk has little to do with deploying the technology. It's the danger of giving up ownership of the decision. That's not something I'll ever let happen, no matter how good any one agent gets at arguing its case.

The Agent for When I'm Not There

Most recently, I built a "CIO" agent: modeled on how I think and what I'd ask, for our executive leadership team and business unit leaders. The goal is greater alignment and more confidence in making decisions without me. It helps leaders prepare and decide with better context, but it is not a proxy for me. The human leader still owns the decision. It also frees me up for the work no agent can do: building the relationships with my executive leadership team that make everything else possible.

I expect our leaders to adopt it readily. The culture doesn't fear more perspective in the room; it expects it.

When I was an executive partner at Gartner, I used to imagine that the hardest CIO job in the world would be leading IT there, surrounded by ten thousand analysts who all thought they knew how to do it better than you. When I took on the CIO role at West Monroe, it turned out to be the opposite, thanks in part to how I framed it. Instead of "they think they know better than me," I started treating that expertise as something to harness. I pushed my own team to find their "doppelganger"—the West Monroe consultants doing for clients what my team does internally—and build real relationships with them. That's where the value showed up: not in avoiding the second-guessing, but in turning it into a resource.

The AI board works the same way. Every agent is there to present a viewpoint or blind spot faster than I'd find it on my own. Building it hasn't replaced my human board. If anything, it's made those conversations sharper. I'm not bringing them my first draft of a thought anymore. I'm bringing them my fifth.

And, to answer the obvious question, the board did weigh in on this article, too, in the end. Zen and Contrarian reviewed a draft, offered five changes between them, and I decided which made it in and how.