The Oscars have a category for actors needed in every movie but never cast as the lead: Best Supporting Actor. The audience knows their faces. The movies depend on them. But they'll never be the star.
The CIO is one of those supporting acts. This is not a complaint about underappreciation. It is a structural observation about the role itself. For most of the last two decades, the CIO's job was framed as a service function. An informed buyer who procures the right systems, integrates and rolls them out without hiccups, and most importantly, always passes the audits.
In many organizations, IT is still seen as "an order taker, not a business strategy driver." Surveys find that only 25% of CIOs feel their work is meaningfully differentiated from competitors, and many describe the role as a "never-ending game of catch-up." Many feel like they don’t get to shape decisions, but have to absorb the consequences.
But that was not always the case. In the 1980s and 1990s, technology leaders ran build shops. Companies wrote their own code, designed their own integrations, and ran their own data centers. The CIO was responsible for the company's strategy in a way that was visible to everyone.
And the ambition is still there. 67% of CIOs say they see themselves as a future CEO (the highest number among CxOs). The supporting cast wants the lead role. And for the first time in a generation, they have a credible path.
IT has been on the wrong side of an unfair equation. The business owned the outcome, whereas IT owned the implementation. When the outcome materialized, the business owned the win. When it did not, IT owned the “implementation gap.”
Agentic AI fundamentally disrespects this separation. AI agents do not just support work, they execute it. This means functionality owned by IT will move money, change records, and make business decisions. So, unless your company’s strategy is to sit out AI and hope for the best, IT will be responsible for a meaningful share of how the business actually runs.
The democratization of AI is also the democratization of risk
With agentic AI, a finance expert can sketch a reporting workflow in an afternoon, and a procurement team can prototype a sourcing assistant in a week. And they do. That is exciting, but it also sets the wrong expectations.
Building something that works once is not the same as building something the enterprise can rely on, or as Bain puts it, it is the difference between episodic agents that show up on demand and long-running agents that carry the work forward over time.
What works for B2C doesn't necessarily work for B2B. You can’t wing it with a new AI agent that touches finance, HR, procurement, supply chains, or compliance. Some companies will try, but I expect many of them to pay the price. Or as Christian Klein recently put it: “While 80% accuracy may be sufficient for consumer AI applications… eighty percent is just not good enough when you run the world’s most business-critical businesses.”
The right path forward is neither to scale AI slop, nor to stop building. The golden path is to turn more AI experimentation into systems that can be trusted, scaled, governed, measured, and connected to business outcomes.
Stop saying no, start engineering yes
This is not just an operational shift, but a mental model shift from owning the how of technology to co-owning the what of the business.
The old caricature of IT was the department of no. The business wants to move fast. IT says no. The business wants new tools. IT says no. The business wants to test something new. IT says architecture reviews, integration plans, governance processes, audit assessments, and a seemingly endless list of variations on the theme of no.
In the era of agentic AI, the CIO cannot be the department of no. CIOs need to become the architects of safe yeses. That means building the infrastructure and processes that let business-led AI run at enterprise scale.
That is a much bigger mandate than technology delivery. And the old mandate hasn’t gotten easier, systems still have to run, projects still have to ship, costs still have to stay under control, and AI makes each of those harder. In an AI-native company, technology delivery and business execution move in lockstep.
This changes the conversation with the CEO and the rest of the C-suite. The CIO should not walk into conversations about who owns agentic AI processes as someone trying to take over. But AI creates a shared execution layer across all of them, and that layer is the one the CIO can shepherd.
The difference is not to say, “this is my process now,” but instead to say, “I can help make this process faster, more measurable, and more valuable by helping you deploy AI correctly.”
The CIO earns a bigger seat at the table by creating the IT backbone that enables seeing what the agents are doing, evaluating whether they are creating value, and adjusting the system when they are not.
Don't count agents, count impact
I have seen a version of this movie before. Many companies pursued robotic process automation (RPA) with high expectations. In theory, it made sense. Automate repetitive tasks, reduce manual effort, and scale the bots. In practice, it turns out that counting bots is not the same as realizing value. Companies ended up with portfolios of bots that increased costs, fragility, and operational complexity at least as fast as they increased impact.
The lesson is not that RPA was useless. The lesson is that delivery metrics become misleading when they are disconnected from value. The fix is not another dashboard. It's value management as a practice. The discipline of tying every deployment to a validated outcome.
The same risk exists with agents. A company can deploy many agents and still not transform. It can automate more activity while creating more cost, more risk, and more noise. It can celebrate the number of agents launched while struggling to explain the value realized.
CIOs must resist the temptation to count how many agents their organization has deployed. Instead, they need to be able to answer the question of how much business impact they created. Did they reduce working capital? Improve throughput? Reduce leakage? Don't count agents, count impact. In the agentic enterprise, IT delivery is value delivery.
CIOs who speak the language of validated business outcomes win. Their unfair advantage is that right now, nobody else has the process observability into where agents are deployed and what they actually do.
The career math is against CIOs
CIOs serve the shortest tenure in the C-suite of about 4.3 years, versus 5.3 years for their peers. Over the past two decades, 85% of S&P 500 CEOs came from just four roles: COO, divisional CEO, CFO, or a leapfrog promotion from below the C-suite. The board picture is no better. In 2025, only 6% of newly appointed US public-company directors had ever served as either CIO, CTO, CDO, or chief AI officer. That’s why some equate the acronym CIO with “career is over.”
But there is a window of opportunity to recast the lead roles in the C-suite. AI reshapes the boundaries within the business. And whoever gets the work done will have a lead role. Not by demanding it. But by building the backbone that lets the business move faster without losing trust.
The backbone needs process reality
By now, the question for any CIO becomes operational. A backbone is only as good as what it can see. If the CIO is going to make enterprise AI safe to scale, the underlying system has to know how work actually flows, where exceptions occur, where value is trapped, and where an agent can act without breaking downstream processes.
This is the business value we want to unlock at SAP Signavio. Process observability turns agent execution into something a business can evaluate and govern, not just deploy. Company Memory gives agents the specific organizational context they need to behave as you intend, rather than the defaults of a generic model. And value management, the discipline I argued for in my previous piece, is the discipline that turns all of it into validated business impact.
The CIO becomes the enterprise value backbone
For twenty years, the CIO has run a procurement organization with a rollout muscle. Before that, the CIO was a builder. AI is the moment the CIO can build again, not by writing code, but by building the backbone the agentic enterprise runs on.
The next phase will not be won by the companies with the most agents. It will be won by the companies whose CIO can make agentic AI safe enough to scale, grounded enough to trust, accessible across the organization, and connected to validated business outcomes. That's how the supporting actor finally gets the lead.
A modest proposal for the “naming-industrial complex”
One prediction I can make with confidence is that a new name for this role will be workshopped. There is a reliable pattern for declaring old titles obsolete and replacing them with a new acronym. Along with a five-level maturity model and paid-for coaching. The rebranding doesn't actually change the role. So, in the interest of saving time, I humbly offer ready-made titles.
- Chief Yes Officer (CYO), for the architect of safe yeses
- Chief Agentic Officer (CAO), until next year's "Post-Agentic Officer"
- Chief Backbone Officer (CBO), finally, a spine in the org chart
- Chief Impact Officer (also CIO), confusion is the point
- Chief Make-It-Executable Officer (C…O), please do not make this an acronym
The honest version is that the CIO title is fine, the mandate is the problem. The role didn't need a new name in the 1990s when the CIO ran build shops, and it doesn't need one now that CIOs can build again. What it needs is the backbone described above. Consultants and thought-leaders will rename it anyway ;)
With thanks to Lukas N.P. Egger for his input, review, and the discussions that helped sharpen this piece.
References:
- Surveys find that only 25% of CIOs feel their work is meaningfully differentiated from competitors, and many describe the role as a “never‑ending game of catch‑up.”
https://blog.adobe.com/en/publish/2019/11/08/study-only-25-of-cios-feel-their-work-is-differentiated-from-c
- Builders → buyers → builders again
https://er.educause.edu/articles/2018/8/a-brief-history-of-the-cio
- 67% of CIOs say they see themselves as future CEO (the highest number among CxOs).
https://www.cio.com/article/4094675/67-of-cios-see-themselves-as-potential-ceos.html
- Building something that works once is not the same as building something the enterprise can rely on, or as Bain puts it, it is the difference between episodic agents that show up on demand and long-running agents that carry the work forward over time.
https://www.bain.com/insights/ais-next-operating-model/
- “While 80% accuracy may be sufficient for consumer AI applications… eighty percent is just not good enough when you run the world’s most business-critical businesses.”
https://news.sap.com/2026/05/sap-sapphire-keynote-business-ai-platform-power-autonomous-enterprise/
- On average, CIOs serve the shortest tenure of about 4.3 years, versus 5.3 for everyone else.
https://www.ciodive.com/news/cios-often-the-youngest-shortest-tenure-members-of-c-suite/436619/
- Over the past two decades, 85% of S&P 500 CEOs came from just four roles: COO, divisional CEO, CFO, or a leapfrog promotion from below the C-suite.
https://www.healthevolution.com/insider/data-dive-which-c-suite-roles-most-commonly-lead-to-ceo/
- The board picture is no better. In 2025, only 6% of newly appointed US public-company directors had ever served as either CIO, CTO, CDO, or chief AI officer.
https://www.heidrick.com/en/insights/boards-governance/board-monitor-us-2026-demographic-report