How Stanford GSB Rewired Its Marketing Team for the AI-Native Era

What Stanford Is Teaching Us About the Work of Going AI-Native 

There is a particular kind of confidence that comes from an institution built to study decisions deciding, for once, to make one about itself. Stanford's Graduate School of Business has spent decades teaching the world how organizations change. This year, its own Dean's Communications team became the case study. 

The team didn't add an AI tool to its stack. It replaced the stack. 

That distinction matters more than it might first appear. Most organizations approach artificial intelligence the way a homeowner approaches a drafty house: they buy a space heater. A chatbot here, a drafting assistant there, a research tool bolted onto whatever workflow already existed. The house stays cold in all the same places; you've just added noise and an extension cord. Kristin Harlan, who leads Dean's Communications at Stanford GSB, and Harjiv Singh, founder of CambrianEdge.ai, both seem to understand this instinctively. Their conversation, recently reported in diginomica, is a shared diagnosis of why so much AI adoption fails to change anything at all. 

Harjiv compares the current moment to the arrival of electricity. You don't wire a building for electricity by hanging a few bulbs from the existing gas fixtures. You rewire the building. The work is slower, less visible in the early going, and considerably more disruptive to the people doing it - which is precisely why most companies skip it and buy the space heater instead. 

He goes further, describing AI adoption as a three-act play. Act One is simply learning to use the tools - the ChatGPT phase, where a workforce discovers what a large language model can do for an email or a slide. Act Two is remodeling the business around what's been learned, which is where most organizations currently sit, uneasily, having outgrown the tool but not yet redesigned the work. Act Three is AI-native operation: a business built, from its first assumptions, around agentic systems rather than retrofitted to accommodate them. Harjiv is candid about what he thinks this often becomes in practice: "I am tired of companies laying off people and blaming AI." The line lands because it is true, and because so few executives are willing to say it in public. 

What makes the Stanford story worth telling isn't the technology. It's the behavior underneath it. Harlan's team - nine people, all willing,they needed to change - had built a perfectly serviceable set of habits. Speeches got written one way. Social posts got written another. When the Dean was asked to speak at an event, the request would scatter across email threads, Slack channels, and whichever tool happened to be open on someone's laptop that morning. None of it was broken, exactly. It simply asked nine talented people to spend their mornings doing the coordination that software should have been doing for them. 

The shift, once it happened, showed up less as a productivity metric than as a change in what people got to spend their attention on. Harlan describes talking points for the Dean flowing into the platform, refined in her own voice, so that the underlying labor of research and drafting no longer competed with the labor of judgment. Video that once required an outside production budget now gets made in-house. And this is the detail worth sitting with - the team stopped cutting and pasting between six browser tabs and started working inside one place built to hold the whole job at once. It is a small thing to describe and, evidently, a large thing to live without. 

Harlan is careful not to oversell the tool as the point. “It is not just a productivity tool,” she says, “it is about figuring out how it will change the way you work.” That is the harder and more honest claim, and it is the one most vendors are reluctant to make, because it requires admitting that the software alone accomplishes very little. It is the willingness to rebuild the process, not merely speed it up, that does the work. Software just makes the new behavior possible. 

There is a broader story here too, one that will feel familiar to anyone who has watched technology enter a large organization through the side door. Marketing teams move fast, sensing an edge before IT has finished its risk assessment. That speed produces real gains and real headaches, a scattering of unsanctioned tools that CIOs discover eventually, usually the hard way. What's different this time is that the platform earning marketing's early trust isn't positioning itself as marketing's tool. It sees the communications team as a foothold, a place to prove the model before extending it to the rest of the enterprise. Stanford, characteristically, seems to have already worked that out for itself. 

The lesson, if there is one, isn't about artificial intelligence at all. It's about the discipline required to finish a renovation instead of just admiring the new light fixtures. Stanford GSB didn't buy a smarter bulb. It rewired the building. That's Act Two. Act Three is still being written, and it's worth watching who has the patience to get there. 

If your marketing team is still coordinating across six different tools, Stanford's Act Two is worth a closer look.