Enterprise AI Will Be Won by Ecosystems - with Sugan Palanee

How do you turn AI experiments into enterprise value? Few people see that question from a better vantage point than Sugan Palanee, Global Chief Operating Officer for Clients and Industries at EY. His role puts him inside the boardrooms of many of the world's largest businesses, right where strategy and change around AI agents are being orchestrated.

In this episode of the HumBot Podcast, host Sanjeev sits down with Sugan to explore why so many AI programs fail to move the needle, what a realistic path to enterprise value looks like, and why the winners of the agentic era will be ecosystems, not single platforms.

A 34-Year Vantage Point on Transformation

Sugan's journey is a story of transformation in itself. He joined EY straight out of school in Durban, South Africa, qualified as a chartered accountant within about five years, and went on to work across the US, the UK, France, and Africa before moving to London to lead business development across Europe, the Middle East, India, and Africa. Today, as Global Chief Operating Officer for Clients and Industries, he oversees EY's largest accounts, its industry go-to-market, and global delivery teams across India, Poland, and Argentina.

Along the way he has lived through the automation wave, the AI wave, and now the agentic wave. And although he is a trained auditor rather than a technologist, he makes a deliberate personal investment in understanding everything tech:

When I talk to clients, I should know more than them.

That mindset, skill up and keep skilling up, runs through the whole conversation. It applies to leaders, to graduates entering the workforce, and to the universities educating them. The workforce EY recruits today is very different from five or six years ago: the lower-level grunt work is increasingly done by AI, and the people coming in are expected to be deeper thinkers, able to integrate, supervise, and manage systems. The open question, in Sugan's view, is whether universities are changing at the pace that commerce and industry now demand.

Stop Agentifying Old Processes

The conversation opens with a sobering statistic referenced from EY's research: force-fit AI onto legacy workflows and the probability of success drops to single digits. The alternative is zero-based workflow design, a complete reimagination of the business workflow in the world of AI and agents.

Sugan's challenge to his clients is direct: don't retrofit AI. Start from outcomes. The outcome you wanted a year or two ago is very different from the outcome you want for the future, so redesign your processes for the answer of tomorrow, not the answer of yesterday.

His practical guidance follows the same logic. Rather than reinventing or reinvesting in ERPs, put an intelligence layer over them. And crucially, quantify what AI actually gives you. AI creates efficiency, but very few organizations rigorously measure the savings.

Stop agentifying old processes. Start designing the new process that gives you a better outcome with less capacity, and quantify what AI really gives you.

Video Highlight: Stop Agentifying Old Processes

The Frozen Middle: Where Change Gets Stuck

One of the sharpest observations in the episode is about where AI adoption stalls inside organizations.

It is rarely the top or the bottom. The C-suite is ready to support. The younger, tech-aware crowd is eager to experiment. The blocker is the middle management layer, which often feels job insecurity with the advent of AI and becomes too scared to truly experiment.

Sugan's advice is to spend more time with lower and middle management: encourage the doers, reward experimentation, and help the middle layer rise with the change rather than resist it. If organizations can make middle management confident about AI, much of the adoption battle is already won.

Can AI Actually Increase Enterprise Value?

Then comes the question that gives the episode its edge. AI has been implemented across the board for three or four years now. So has it improved enterprise value?

Sugan runs the thought experiment: if all the listed companies across the globe had successfully adopted the AI agenda, they should be improving their enterprise value by at least 20 to 25 percent. Yet he has still not seen a corporate that can clearly articulate an increase in enterprise value attributable to AI.

If AI adoption were truly working, enterprise value should be up 20 to 25 percent. Nobody has quantified that yet.

Why the gap? Some companies have implemented AI improperly. Others have implemented it but never realized the savings, keeping the same cost base instead of evolving people toward bigger, more strategic work. And many have invested without the requisite ROI.

For Sugan, this is the opportunity. Work backwards from a clear end-game: define the delta in enterprise value you want to achieve, break it down into components, and be honest that some of those components might not be AI at all. To the extent they are, let your people innovate and let your ecosystem challenge what you have, so that the sum of the parts adds up to the whole.

Video Highlight: Can AI Actually Increase Enterprise Value?

Governance at the Core, Innovation at the Edges

How do you keep control without killing creativity? Sugan's answer mirrors how EY runs its own global organization: governance is developed centrally, at the global level, while innovation is actively encouraged locally.

Country teams, service lines, and verticals are closest to unique practices and unique problems. Give them room to think out of the box, then pick up the best and brightest ideas, evolve them, and industrialize them into global standards. Stifle innovation at the local level and you never get the best of your organization.

The same logic applies to shadow AI teams. You cannot kill innovation, but you can insist on belt and braces at the central level: clear guardrails, strong governance, and leadership that makes sure nobody blurs the lines. Innovation should be rewarded, not stifled, and always within the guardrails.

Three Questions Before Scaling AI Agents

For an enterprise executive wrestling with how to execute agents at scale, and how to drive real enterprise value change rather than small victories, Sugan offers three areas to think about and act on:

1. Is your data clean enough? Everything else builds on this foundation.
2. Are you over-reliant on one platform? Don't assume the solution sits with a single provider. There are excellent players in the market, each with a unique strength. The question is how you orchestrate that ecosystem to get the best value.
3. Are you empowering your own people? Your people know your systems better than any service provider ever will. Encourage them to innovate and to design for the outcomes of tomorrow, then bring in service providers as a clean pair of eyes to assess and orchestrate.

The summary is memorable: it's data, it's ecosystems, and it's people.

Video Highlight: Three Questions Before Scaling AI Agents

The 80/20 Inversion: People Still Make the Delta

AI is not replacing people, Sugan argues. It is changing what people do. The traditional 80/20 of the workforce is inverting: rather than most people doing the work, most people will coordinate, orchestrate, and manage systems, and someone has to manage the machines. You can only manage a machine if you understand how it works.

That creates an obligation for employers. Whether people stay or move on, businesses have a moral duty to empower them with the knowledge to orchestrate. EY takes this seriously: through internal programs such as EY University, everyone, including partners like Sugan himself, goes through pass-or-fail AI courses to stay up to speed.

The link back to enterprise value is direct:

Your enterprise value is not going to change if you are over-reliant on machines. Your people are what make you, not the machines.

Even as pricing shifts toward outcomes, it is people who deliver those outcomes, and people who are the reason outcomes get delivered or not.

Pace, Budgets, and the Token-Cost Wake-Up Call

On transformation roadmaps, Sugan is pragmatic: work at the pace the client is comfortable with, not the pace that suits the service provider. In an environment of constrained budgets and global uncertainty, that increasingly means breaking large transformation programs into smaller pockets that are easier to manage, and avoiding major ERP changes in favor of using AI as the intelligence layer on top.

He also flags a cost dynamic that has become a burning platform for clients: token costs. The most innovative layer of the workforce, the level below middle management, is going wild on experimentation, and the token costs add up fast. The answer is not to stop the innovation but to manage it: budget for it, negotiate with the different players for the best value, and keep experimentation from becoming a strain on the income statement.

One Neck to Choke? Why Ecosystems Win

The theme that gives the episode its title emerges from a challenge one of Sugan's clients put to him: "I want one neck to choke."

It is a fair ask, but as Sugan puts it, the one neck you choke might be the wrong neck. The market is now so broad, and the depth of experience so vast, that he struggles to find a single organization that can deliver A to Z off an enterprise shopping list. Ecosystems are the future. You can still have an anchor, and that anchor does the orchestration, but the value comes from bringing the best of the ecosystem together.

EY has made this deliberate: an alliance leader sits on the global executive, orchestrating formal alliances with players such as SAP, Microsoft, and Snowflake, and going to market together. The point is not one firm winning the trophy or capturing the entire fee. It is delivering the best service to the client, not a substandard one born of selfishness.

This is also where deep specialists fit in. As Sugan puts it, this is what creates opportunities for the HumBots of this world: unique, deep, cost-effective capability with the client at the forefront of everything they do. Multiply that by ten and you have the ecosystem clients actually want to buy, whether they are global multinationals or middle-market organizations.

Ecosystems are here to stay. That is the reality.

Video Highlight: Enterprise AI Will Be Won by Ecosystems

When Agents Act, Humans Stay Accountable

In the last year, agents have moved from advising to acting. With the right data and integrations, they can execute, which makes authorization a board-level question.

Sugan's position is unambiguous: you will never allow an agent to operate unilaterally on its own. Humans provide the governance. They monitor inputs and outcomes, and everything an agent does should be reviewed and signed off by a human with the depth of experience to manage it.

That is a message of empowerment, not restriction. The accountability lies with the human; the operationalization lies with the bot. People gain more authority in this model, and with it more accountability, which is exactly why their jobs are not at risk if they evolve to remain relevant.

The Real Challenge Isn't AI. It's the Human Factor.

Asked what learnings from previous transformation waves still apply, Sugan's answer cuts through the technology hype:

I don't think your challenge is AI. I think the challenge is the human factor.

AI is here to stay. The hard part is the change of mindset: empowering people so they are not filled with the fear of being replaced, but understand they are being empowered, to make decisions they were never able to make before, on outcomes that machines now generate.

The critical conversion is that middle layer that never studied this technology and now occupies key positions. Get them confident about AI, and, in Sugan's words, by and large the world will be very successful.

Final Takeaway

The closing summary of the conversation is a simple sequence: data, ecosystems, and people.

Data is potentially the easiest part. Ecosystems come next, orchestrating the best players rather than betting everything on one platform. And the biggest change, the one that determines whether AI ever shows up in enterprise value, is on the human side.

The enterprises that win the agentic era will be the ones that redesign processes for tomorrow's outcomes, orchestrate ecosystems instead of chasing a single neck to choke, and empower their people to manage the machines.