Want to build an AI operating model? First, build consensus



[

At a recent dinner with chief operating officers in New York City, we caught up with the senior leaders most often tasked with transforming AI use into value. The conversation was markedly different from the ones we were having just one year ago.

In 2025, peers in the C-suite would boast about individual AI successes: An agent that made software engineers more productive, or an AI use case in accounting that dethroned the spreadsheet.

But this year, conversations were focused on the forest, not the trees. Individual successes felt insufficient. The bigger question was: How do you do this again and again? How do you build a foundation that makes AI success second nature?

Leaders want an AI Operating Model: an enterprise-wide foundation that makes agents, automation, and intelligent data routine. Because the alternative is exhausting: fighting tooth and nail to scale individual AI projects.

Unfortunately, exhaustion is the status quo.

Why? One reason is a misconception about who builds the AI Operating Model. Ask who owns AI, and every executive raises a hand: the CIO for the technology, the CAIO for the models, the CFO for the investment, the CHRO for the workforce.

In reality, it’s a joint effort across the C-suite. Still, someone must steward these efforts. Often that’s the COO, whose remit spans every function. But even the COO can’t make executive decisions across every business unit. Instead, they need to build consensus. And so a decidedly old-school skill – influence – is at the center of AI success today. Rolling up your sleeves and building relationships, not AI models, is what matters most.

Let’s start with the AI Operating Model. What is it, exactly? An AI Operating Model is a reusable foundation underneath all your individual AI investments. At its core, the operating model connects four key areas. First is intelligence: trusted, real-time context that lets agents make smart decisions. The second is action: The ability for that intelligence to span the enterprise, from people and workflows to applications and APIs. Then comes operations – the ability to observe and optimize that action across the business. And finally, control: Ensuring AI has rigorous governance and security.

Intelligence, action, operations, and control span front- and back-office functions. They touch technical and non-technical roles. They involve employees at all levels. This is where consensus building comes in. Being able to shape decisions, rather than execute them with complete authority, can be more valuable. The real work is building consensus into how the enterprise runs, so it holds from one initiative to the next.

When we sit down with clients, that last step is where we see most companies stall. Leaders agree on the vision, then each function goes off and builds its own AI, and the agreement never makes it into the work. The technology is ready: Taking a company’s standard operating procedures and turning them into agentic use cases, an ROI case, and a blueprint used to take days or weeks. Now it can take hours. The hard part is agreeing on what to build. The symptoms show up in three places. On value, CEOs report that only 25% of AI initiatives have delivered expected ROI over the last few years. On workflows, 82% of C-suite executives say functional silos block value. On platforms, half of CEOs say the pace of recent investments has left their organization with disconnected, piecemeal technology. Closing those gaps takes consensus.

We didn’t get this right at IBM from the start. Our early wins came quickly, as they do for most companies. What has let us keep scaling was consensus. We now apply the same practices with clients, and start by asking a few simple questions:

  1. Are you eliminating work before we automate it? At IBM, every workflow went through the same sequence: eliminate, simplify, then automate. Automating a broken process only scales the inefficiency. The order works because our CEO and leadership team agreed first on where AI would create the most business value.
  2. Are you redesigning workflows, or just automating tasks? IBM chose end-to-end workflows within each domain, not isolated tasks, and turned them into funded, phased plans with the business bought in. Small cross-functional teams, often a handful of people working alongside AI agents, carried each one through. Our methods are built into the workflows, agents, and assets those teams use, so each initiative starts where the last one left off.
  3. Can everyone see who owns which result? Every project and domain rolls up to a single enterprise scorecard, with leading indicators such as adoption that show what is working before savings hit the ledger. It is how a $2 billion productivity goal became $4.5 billion in savings in three years. We now bring the same approach to clients.

Today’s AI leaders must be savvy technologists, but that’s not enough. They also need the necessary credibility, communication, and consensus-building skills. The most impactful AI leaders excel when they surface evidence to shape decisions about AI investments.

If AI leaders can’t do this, they’ll launch one-off AI projects and chase ROI. But if AI leaders can get peers on the same page about an AI Operating Model, they’ll build an AI value creation flywheel. One where productivity fuels investment, and investment fuels greater growth.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.

https://fortune.com/img-assets/wp-content/uploads/2026/11/GettyImages-2277251072.jpg?resize=1200,600
https://fortune.com/2026/11/19/ai-operating-model-build-consensus-ibm/


Joanne Wright, Yogi Goyal

Latest articles

spot_imgspot_img

Related articles

Leave a reply

Please enter your comment!
Please enter your name here

spot_imgspot_img