The Month I Became My Company's De Facto AI Strategy Lead (Without Anyone Asking Me To)

We were near the bottom of a long Tuesday agenda when our CEO looked up from his laptop and asked the room a question that wasn't on it: "Who's actually responsible for making sure AI is being used across the company — not piloted, not talked about, but used?"

Then came the silence. I counted it later, replaying the moment: about ten seconds. Our CTO looked at our Head of People. Our Head of People looked at our VP of Sales. Our VP of Sales looked at his coffee. Nobody said the obvious thing, which was that there was no answer, because there was no owner. And then — and I felt this happen before I understood it — the room's attention drifted toward me. Not because anyone had decided I should own it. Because I was the only person at the table who could have told you, function by function, what every team was and wasn't doing with AI. The sales team's note-taker experiment. The half-built support deflection bot. The two product managers quietly running their roadmaps through a model nobody had vetted.

I didn't volunteer. I just looked back at him and said, "I can take a first pass." Four words. By the time I'd finished the sentence, something had been decided that no one had actually decided. I had a new job. Nobody wrote it down.

The Short Answer

AI is changing the Chief of Staff role in a specific and underappreciated way: it's turning the CoS into the de facto owner of how AI gets used across the organization — without anyone making that decision formally. Because the CoS is the only person who sees across every function, knows where context lives, and can identify where AI is actually helping versus where it's being piloted without accountability, they become the natural governance point. Most CoS leaders are already doing this. Almost none have a job description that says so.

Why Nobody Assigns This Role (And Why It Lands on the CoS Anyway)

Here is the structural reason the room looked at me. AI ownership has no natural home in a standard org chart. IT owns the infrastructure and the security review. The CTO owns the AI inside the product. Whoever runs People owns the rollout of any tool that touches performance or hiring. Each of those owners is real, and each of them owns a slice. None of them owns the whole.

The whole is the part that matters and the part nobody is accountable for: is AI actually changing how we operate, across every function, or are we just accumulating a drawer full of pilots that each made sense in isolation? That question doesn't belong to a function. It belongs to the seam between functions. And the seam between functions is where the Chief of Staff has always lived.

This is not a new job. It's the oldest job I have, pointed at a new domain. Long before AI was the question, I was the person who could tell the CEO that the thing Sales believed about the product roadmap and the thing Product believed about the sales pipeline were two different things — and that the gap was costing us deals. Synthesizing cross-functional reality is the work. AI governance is just the newest place that work is urgently needed.

What makes it land on the CoS rather than on a committee is speed. The CEO didn't want a task force. He wanted to stop feeling like the company was doing something important without anyone watching it — the same instinct that makes a leadership team want a single, current picture of where everything stands instead of a status meeting that produces one. The person who can give him that picture fastest is the person who already holds the map.

So if you're the one who holds the map, the question isn't whether you'll end up doing this. You will. The only real choice is whether it accretes onto you invisibly until you're accountable for something nobody agreed you owned — or whether you make it deliberate, with your CEO's backing. I'll come back to exactly how I had that conversation. First, what the job actually turned into.

What Owning the AI Agenda Actually Looks Like

I want to be honest about the gap between using AI and owning it, because it's wider than it looks. The research that finally gave me language for it comes from a 2026 study of Chiefs of Staff: 86% of us use AI daily, but only 7.3% are what the study called "AI-Native" — operating with systems where AI runs continuously rather than being summoned task by task (chiefofstaff.network, 2026). I was firmly in the 86%. I drafted with AI, summarized with AI, cleaned up my own messy thinking with AI. None of that is ownership. Ownership is the 7.3% problem.

Owning the AI agenda meant I stopped asking "how can AI help me get through my week" and started asking "where is AI quietly working, where is it quietly failing, and who would even know?" The second question is governance. And governance, it turns out, is mostly a visibility problem before it's a policy problem.

The silent failures were the ones that scared me. A pilot that fails loudly gets killed and learned from. A pilot that fails quietly just erodes trust while everyone politely stops using it — the support team that drifts away from the deflection bot after it gave a customer a wrong answer in March, and never says so out loud. I needed to see the slipping thing on day three, not day thirty. That's the exact problem we built Rhythms' Radar to solve: it scans the connected tools and surfaces what's going sideways before anyone has to go looking, which for AI initiatives matters double, because the quiet failures are the ones that poison the well for the next attempt.

The hardest part wasn't technical. In a 2026 survey of operators, more than nine in ten cited cultural challenges — not technology — as the principal hurdle to AI adoption (founderoperator.com, 2026). That matched my month exactly. The blocked experiments weren't blocked because the models didn't work. They were blocked because nobody trusted an output they couldn't trace, because one team was afraid another would judge their use of a bot, because a manager had quietly decided AI was a threat and slow-walked it. You don't fix that with a procurement decision. You fix it the way a Chief of Staff fixes everything: by being the trusted neutral party who can see the whole field and tell people the truth about it. That's also why this can't sit with IT — IT can tell you whether a tool is secure, but not whether a team has quietly stopped trusting it.

The Three Things I Now Track Across Every Function's AI Use

By the end of the month I'd stopped trying to track everything and narrowed it to three things. These are the three I'd hand to any CoS who just inherited this without a mandate.

The first is where the operating cadence itself is run by AI versus merely assisted by it. Most teams use AI to draft a thing a human still has to assemble. The higher-value pattern is AI preparing the recurring work — the weekly review, the pipeline check, the customer health read — from live data, so the human walks in to judge rather than to build. It's the difference I see between the teams getting real leverage and the teams with the most clever prompts. This is the footing we put our own leadership cadence on with Rhythms' Reviews: the pre-read assembles itself from the connected systems, so the meeting starts with reality instead of three people reconciling spreadsheets.

The second is where AI is failing silently — the day-three signal I mentioned earlier. I keep a running read of which experiments are losing trust before anyone formally pulls the plug, because a quietly abandoned pilot tells you more about your culture's readiness than a successful one does.

The third is whether decisions about AI are actually carried forward. We'd decide something about a tool in one meeting and re-litigate it three weeks later because nobody remembered the decision or the reasoning. The fix is treating AI governance the way good operating systems treat any priority: as something that cascades and updates rather than something that gets re-explained. When a decision changes at the top, the downstream owners should see it without me running a re-communication tour — the same logic Rhythms' Goals & Alignment applies to any company priority. AI governance is just a priority that happens to be new.

None of this required a title. It required a system for seeing across the company — which, not coincidentally, is the thing the CoS role has always quietly needed and rarely had.

What I'd Tell the Chief of Staff Who Just Had Their Own Ten-Second Pause

So here's the conversation I promised. If you're reading this because you just lived your own version of the Tuesday meeting, the thing I wish someone had told me in the silence is that the question of whether to claim this is already answered. You're going to end up doing it either way. The only thing you control is whether you do it deliberately, with explicit backing, or whether it lands on you invisibly until you're accountable for something nobody agreed you owned.

So have the conversation. Not "should I own AI strategy" — that sounds like a power grab and your CEO will hear it as one. Say instead: "There's no single owner for whether AI is actually working across the company, and I'm already the closest thing to it. I'd like to make that real, with your backing." That reframes a quiet inevitability as a deliberate decision, which is the only version that comes with the authority to actually do the job.

And then build the visibility before you build the policy. The reason this is a CoS job and not an IT job is that the difficulty is overwhelmingly cultural, and culture is read, not configured. That reading takes time you probably don't have — the "up to 70% of management time consumed by work about work" is real, and you won't free up the hours for AI governance unless you've already pulled the review prep and status chasing off your plate. Getting the lower-value coordination automated isn't a tangent to this role. It's the precondition for it.

The risk of stepping back because it's not in your job description is that someone else steps forward with a fraction of your context. The most strategically important emerging function in the company should not be run by the person with the narrowest view of it. And right now, the person with the widest view is almost certainly you.

I still don't have a title that mentions AI. I'm not sure I need one. What changed after that Tuesday wasn't my business card — it was my understanding of what the job had quietly become. The Chief of Staff has always been the person who makes the invisible legible for the CEO: the cross-functional truth nobody else can assemble. AI didn't change that job. It just handed it the most important thing it's ever had to make legible. The ten-second pause wasn't the room failing to find an owner. It was the room finding one. It just took me a beat longer than everyone else to notice it was me.

If you're the one your leadership team looks at in the silence, try Rhythms for free at rhythms.ai — and see what it looks like when the operating layer runs itself, so you can do the synthesis only you can do.

Frequently Asked Questions

How is AI changing the Chief of Staff role in 2026?

The biggest change isn't the tools — it's the ownership vacuum. AI strategy in most organizations has no clear functional home: it's not HR, not IT, not the CTO's product work. Chiefs of Staff are filling this vacuum by default because they're the only people who see across every function. In practice, the CoS role is becoming the AI governance and implementation-coordination role — usually without a title change, a budget, or a formal mandate. If you're already the person the room looks at when AI ownership comes up, the change has already reached you.

Should a Chief of Staff own AI strategy at a company?

In most mid-to-large companies, yes — but deliberately, not by accident. The CoS already holds the cross-functional view that AI governance requires, and the alternative is that ownership accretes invisibly or lands with someone who can only see one function. The move is to make it explicit with the CEO: name that there's no single owner for whether AI is actually working across the company, and ask to make that real with their backing. Owning it without that backing means accountability without authority, which helps no one.

How does a Chief of Staff demonstrate value in an AI-first organization?

By using AI for operational visibility, not personal productivity. Drafting emails and scheduling faster is fine, but it's invisible work that nobody sees. The high-value move is generating the leadership pre-read from live data, surfacing off-track initiatives before anyone has to go looking, and maintaining a decision log that makes the CEO's operating cadence self-documenting. When the CoS uses AI to make the operating layer visible, the CEO stops asking "how are things going?" because they already know — and the source of that clarity is unmistakably you.

What AI tools should Chiefs of Staff be using in 2026?

The most durable CoS AI stack has three layers: a review-preparation layer that pulls live data from connected tools before leadership meetings, a decision-tracking layer that captures and carries forward decisions and ownership between sessions, and a signal-monitoring layer that flags off-track initiatives before you have to chase them. The specific products matter less than the architecture. Ask the real question: are all three layers covered, or have you just bolted a clever assistant onto one of them?