Why the founder of OpenExO tells boards that the AI question is not which tools to buy, but whether they still need the machine they are speeding up.
Take a large company and ask what most of its structure is for. Layers of management, meetings, reports, approvals, a budgeting cycle: an apparatus whose job is to move information around and get people to act on it together. That design made sense when intelligence was scarce, information moved slowly and coordinating people was expensive.
Salim Ismail’s argument is that all three of those conditions are now collapsing at the same time, and that most companies are responding by pointing the new technology at the old architecture. “Most companies today are using AI the way we used the internet in 1996,” he wrote this month. “A website here, an email there. Lots of experiments, very little transformation.” The survey numbers agree with him. In McKinsey’s 2025 global survey on the state of AI, only 39 percent of respondents reported any effect on earnings at the enterprise level, and most of those put it below five percent of EBIT.
Salim is the founder and chairman of OpenExO, the author of Exponential Organizations and Exponential Transformation, the founding executive director of Singularity University, and, since January, co-founder and partner at Exponential Venture Capital. What he offers boards and leadership teams is a different first question. Not how AI fits the company, but what the company would look like if it were built around AI.
“If you were starting this company today, with today’s technology, would you design anything remotely resembling the organization you currently run?”
Salim Ismail, founder and chairman, OpenExO
“Usually the answer is no,” he says. “And that’s the scary bit.”
Six smart people and the machinery
He traces the realization to Yahoo, around the period he ran Brickhouse, the company’s internal incubator. “Yahoo had extraordinary engineers. Extraordinary product people. Plenty of capital. Plenty of ideas. None of those things were the problem,” he says. “The problem was the organization.”
“You could have six smart people in a room who knew exactly what should be built and could probably prototype it in a few weeks.” Then the idea would hit the machinery of a large company, “and suddenly the technology was the easy part.”
“We tend to think innovation is constrained by ideas or technology,” he says. “Very often it isn’t. The constraint is the cost of getting humans to agree, coordinate and act.”
Brickhouse was the attempt to step outside that machinery: a small team, given room, building without asking the whole organism for permission. The idea became, in his words, how Singularity University was built and “a huge part of the ExO model.” “The breakthrough wasn’t ‘small teams are better.’ It was realizing just how much organizational energy gets consumed simply coordinating the organization with itself.”
He had met the same shape of problem before Yahoo, in a very different place.
A bridge after 9/11
After the attacks of September 2001, federal money was available to help New York businesses recover. “The money existed. The need existed,” Salim says. “But the bureaucracy between the two was so painful that the resource effectively couldn’t reach the people who needed it.” He started the New York Grant Company to sit between the two. “So we built a bridge.” By his account the company served hundreds of clients and helped move millions of dollars of federal funding into the local economy.
Looking back, he sees a pattern. “I’m almost always trying to find the constraint and remove it rather than optimize around it.” Yahoo was, in his words, a completely different world and “weirdly the same problem.” “So again: build a bridge around the constraint.”
That became, he says, one of the deepest lessons in ExO. “Don’t attack the immune system of the core organization head-on. Build something at the edge. Let it operate under different rules. Show that it works. Then let the evidence infect the core.” It is, he adds, much easier than “asking a 100-year-old organism to voluntarily dismantle the structures that made it successful.”
The edge is where the ExO Sprint does its work.
Twenty-five people, ten weeks
The ExO Sprint is the method OpenExO runs inside organizations. What Salim likes about it is the question it opens with, one he says normal strategic planning almost never asks. “Not: ‘How do we get 10% better?’ But: ‘How would we get 10X better?’ Because the moment you ask for 10X, you can’t optimize the existing system anymore. The existing system wasn’t designed to produce that result.”
His strongest example is Procter & Gamble, which he describes as dealing with sluggishness in its shared services division. Instead of the incremental response, 25 people were pulled together over 10 weeks and made to work backwards from a radically different outcome. The questions change immediately. “You stop asking, ‘What assets do we own?’ and start asking, ‘What assets could we access?’ You stop asking, ‘Who can we hire?’ and start asking, ‘What community could we mobilize?’ You stop asking, ‘How do we get the hierarchy to move faster?’ and start asking, ‘How much of the hierarchy do we actually need?’”
One year after the Sprint, according to P&G, about 20 projects had been launched that yielded about a billion dollars of measurable benefit. “For me the most important result is not the number. It’s the psychological shift.”
“People suddenly realize: ‘Oh. The constraint wasn’t reality. The constraint was the way we had organized ourselves.’ Once people see that, it’s very hard to unsee it.”
Salim, founder and chairman, OpenExO
That shift is also where the strongest objection to his argument begins.
Where the corporation does not disappear
Asked who disagrees most strongly with the coordination-cost argument, Salim doesn’t nominate a person. He gives the objection its best form instead. “It says: companies don’t only exist because coordination is expensive. Organizations also create trust. They allocate capital. They establish accountability. They resolve disputes. They manage liability. Somebody ultimately has to be responsible when the nuclear plant explodes. AI doesn’t make those things disappear.”
“And I agree,” he says. “Where I differ is that we’ve bundled governance, management and coordination together as though they’re the same thing. They’re not.” Governance, accountability, purpose and boundaries all remain, in his view, important. “But do you need seven layers of management to move information from the bottom of an organization to the top and then move a decision back down again? Probably not.”
So the corporation, he says, does not disappear. “I think management gets unbundled from governance. And that’s a much bigger change than most people realize.”
It is also a bigger change than he was describing five years ago.
Why are there 14 steps?
The line Salim finds himself repeating, to clients and to his own team, is the one in the headline. “Don’t use AI to automate the organization you already have. Redesign the organization around what AI now makes possible.” The temptation with every new technology, he says, is to apply it to the old architecture. “We electrified factories but initially kept designing them around steam engines. We put computers into companies and initially used them to automate paper. We’re doing exactly the same thing with AI.”
“If your AI strategy is basically, ‘Take our existing 14-step process and make each step 20% faster,’ you’ve missed the point. The interesting question is: why are there 14 steps?”
It marks what he has changed his mind about. Five years ago he would have said the challenge was making organizations dramatically more agile. “I don’t think that anymore. I think that question is increasingly too timid.” “If an AI agent can do in thirty seconds what previously required information to move through four departments, three meetings and two approval committees, optimizing the meetings is not the answer. You eliminate the meetings.” He has moved from thinking about organizational agility to thinking about organizational first principles. “Don’t accelerate the old machine. Ask whether you still need the machine.”
The habit of asking what happens when a parameter changes, rather than what a technology does today, he credits to Ray Kurzweil. “Human beings are terrible at intuitively understanding exponential change.” The question he took from it: “What happens when it is ten times better, ten times cheaper and available to everyone?” It has influenced more of his work, he says, than almost anything else.
Being right about where technology was heading has not always meant being right about when.
Early is not the same as right
Of the hard stretches of his career, Salim names PubSub as probably one of the hardest. The company was very early. “We could see a world of real-time information streams before that idea had really become mainstream, and intellectually I remain proud of what we saw. But being early and being right are not remotely the same thing as building a successful company.”
“That period was exhausting. It consumed enormous emotional energy, time and attention. When you’re a founder, a company doesn’t stay at the office. You carry it around in your head 24 hours a day. And when things aren’t working, you internalize that.”
“Timing is not a footnote in entrepreneurship. Timing is part of the product.”
Salim, founder and chairman, OpenExO
It also made him, he says, much less romantic about startups. “People tell the glorious founder story afterwards. While you’re actually living it, much of it is uncertainty, stress and repeatedly finding out that reality has absolutely no obligation to cooperate with your PowerPoint.”
He is as direct about the career advice he was given. His father pushed him to pick a lane: specialize, build expertise in one domain, follow the ladder. “That model worked quite well in a relatively stable world,” Salim says. “But almost everything interesting in my career has happened at the intersections between fields.” His estimate is that the half-life of a narrow skill used to be about 30 years and is now about three. The advice he would give now is nearly the opposite. “Get very good at connecting things that normally live in different rooms. That’s where a lot of the breakthroughs come from.”
Where Salim Is Now
In January Salim co-founded Exponential Venture Capital, which also goes by The Organizational Singularity Fund, backing founders building what he calls exponential growth companies. He remains chairman of OpenExO, publishes a weekly newsletter, The Shift, and runs a monthly workshop called The 10X Shift. He gives more than 150 talks a year, this month including the inaugural DBS Asia Future Forward Conference in Singapore. He has also been running a thought experiment with leadership teams: if you could hand one entire job in your organization to AI tomorrow, which would it be? “The room usually goes quiet,” he wrote. “Then someone names a job, and someone else gets uncomfortable. That discomfort is the organisational immune system waking up.”
He doesn’t see Exponential Organizations, ExO 2.0 and what he now calls ExO 3.0 as three different ideas. “They’re really three snapshots of an accelerating technological environment.” AI, he says, takes the work somewhere more fundamental. For roughly a century, companies have been designed around three assumptions: intelligence is scarce, coordination is expensive and information moves relatively slowly. “Now all three are collapsing at the same time.” The outcome, in his view, isn’t a more efficient twentieth-century corporation. “It may be an entirely different species of organization.” That is what he means by the Organizational Singularity: the point at which the organization on the other side of the change barely resembles the one that went in.
The people he wants to hear from are boards and leadership teams asking what it will take to stay relevant in 2026 and beyond.
Asked what he would be doing if he had never gone into this field, he says physics, which is where he started. “I still love the fundamental question behind it: what the hell is actually going on here? How does reality work?” In some ways, he says, he is still doing physics, with organizations instead of particles. “You observe a system. You look for the underlying rules. You try to identify the constraint. Then you ask what happens when one of the fundamental parameters changes. And right now the parameter changing is intelligence. That’s a pretty interesting experiment to be alive for.”
Five Lines from Salim’s Diary
•“Don’t accelerate the old machine. Ask whether you still need the machine.”
•“Timing is not a footnote in entrepreneurship. Timing is part of the product.”
•“Get very good at connecting things that normally live in different rooms.”
•“I think management gets unbundled from governance.”
•“The constraint wasn’t reality. The constraint was the way we had organized ourselves.”
From the Diary of Salim Ismail
Salim Ismail is a United States-based futurist, author and investor, and the founder and chairman of OpenExO.
| Role | Founder and Chairman, OpenExO. Co-Founder and Partner, Exponential Venture Capital |
| Based in | United States |
| Previously with | Yahoo (Brickhouse) and Singularity University (Founding Executive Director) |
| Also with | Board of the XPRIZE Foundation |
“Don’t use AI to automate the organization you already have. Redesign the organization around what AI now makes possible.”
Editor’s note
Executives Diary features Salim Ismail because his argument about AI is an argument about organizations, and he has been making a version of it since the problem had nothing to do with AI. The record runs from a grants company built around New York’s post-9/11 bureaucracy, through Yahoo’s Brickhouse and Singularity University, to a method that has been run inside Procter & Gamble and a fund that invests on the same thesis. What convinced the editors is that he gives the counterargument its full strength before answering it, and that he counts the cost of being early as plainly as he counts the benefits of being right.
Executives Diary, Editorial



