How leaders move AI from proof-of-concept to proof-of-value

By Vishy Narayanan and Sau Shiung Yap

When we talk with business leaders about AI, the first question we’re usually asked is: “Can we start with a proof-of-concept?” It’s a fair question. But we’ll usually pause and ask another one in return: “What are you actually trying to prove?” There’s little need to prove that Copilot can write minutes or that a GPT model can summarise documents. These capabilities are the baseline and they’re improving every day. The real question is: How do you turn this into value for your organisation? So rather than spending time validating what’s already well understood, without clear returns, we encourage clients to shift from proof-of-concept to proof-of-value.

Proof-of-concept is isolated—testing a narrow use case in a sandbox.

Proof-of-value is embedded—wiring AI into real value chains, into real processes, with your data, culture and people.

If you embed AI across the value chain, not just tested in a pilot, you see real transformation. It’s no longer about where to point AI to tweak a process. It becomes about using AI to reinvent the process altogether. You move from doing things differently to doing different things. This is where we see the commercial impact start to show.

This article unpacks how leaders can move from proof-of-concept to transformative value, and where we are seeing it work in practice.

Tools don’t change businesses. Leaders do.

Right now, many organisations are spending heavily on AI tools and technology. Boards are inundated with media headlines promising dramatic productivity gains and new revenue from AI and put pressure on management teams to deliver it. Yet 50% of CEOs report seeing little to no financial upside of using AI, according to our recent 29th Global CEO Survey—Asia Pacific. That gap between promise and reality is where the challenge sits. So what’s driving it?

Most of the AI adoption we see is focused where the pain is highest. This is often data-intensive work that drains employee time and energy, time-intensive processes with long cycle times, or tasks that are prone to error and frequent rework. Too few CEOs are focused on rewiring their organisations to use AI, reshaping strategy, culture, roles and decision-making to make the technology valuable. This is backed up by data. Only 26% of organisations report having strong foundations across at least six of seven core areas, including culture, strategy and AI roadmap, Responsible AI, talent, investment, technology environment, and data, according to our 29th Global CEO Survey—Asia Pacific. Yet those with stronger AI foundations are twice as likely to see both revenue growth and cost reduction from AI.

Organisations must invest as much in transforming how the business runs, and in supporting their leaders through that change, as they do in the technology itself.

“Technology alone won’t deliver value. Organisations much invest equally in leadership along with the tools.”

As Microsoft CEO Satya Nadella recently cautioned at Davos, “for this not to be a bubble, by definition, it requires that the benefits of this technology are much more evenly spread.”¹ In other words, if AI value stays concentrated with technology firms or in house technology teams, the promise won’t hold. Value only materialises when organisations rewire how work gets done, and it delivers real, widespread value across industries.

In a further challenge to the status-quo, Ieading an AI-driven organisation today means leading a workforce that is no longer entirely human. It demands new leadership capabilities.

“Today’s CEOs are the last to manage human workforces.”

Marc Benioff, CEO of Salesforce

We see five leadership shifts that matter most:

1. AI mindset: From awareness to AI fluency

Leaders don’t need to be technologists, but they do need a practical understanding of what AI can do for the business. It’s a move from do as I say to do as I do: trying the tools, modelling experimentation and showing curiosity along the way.

Crucially, this helps enable the necessary mindset shift from doing things differently to doing different things. Our landmark global Value in Motion study shows how AI, alongside other global forces, is creating new growth domains. It’s enabling new business models. Forging new markets. Shaping new customer needs and preferences. To capture this opportunity, companies need to be willing to reinvent themselves. By 2035, there’s an estimated $49.99tn of value at stake across Asia Pacific for those that do.

Stripe, the global payment processing platform, offers a clear example of this mindset in action.² AI and machine learning are embedded into the core of its payments platform, from fraud detection to authorisation optimisation, and are used daily to inform decisions at scale. Rather than treating AI as a specialist capability, Stripe’s leaders have built it into how the business operates and learns. This shift from awareness to practical fluency enables better decision-making, and continuous improvement across the platform. As Stripe’s footprint grows across Asia Pacific, that fluency is increasingly important in fast-scaling markets.

2. Vision and strategy: From FOMO to focus

Organisations that scale AI successfully are clear about what they are building towards, not just what they are experimenting with. Without that clarity, AI can quickly become a game of FOMO rather than focus. Clarity starts with the right questions. How might AI change what your business is, not just how it runs? And as new AI-enabled growth domains emerge, where could your business play—and where could it be left behind?

This shift to strategic intent is visible with the emergence of ‘physical AI’: systems that don’t just analyse data or generate insights on a screen, but act in the physical world. NVIDIA CEO Jensen Huang has been explicit about where this is heading, “every single car will be AI-powered in the future. Every car company will have two factories: one for building cars, and one for building the AI that runs them.” ³

This isn’t about running more pilots. It’s about setting a clear strategic direction for how AI becomes integral to products, operations and infrastructure.

We’re seeing this vision translate into action elsewhere, too. Manufacturers and heavy-industry players are collaborating on autonomous fleets and AI-assisted machinery, while advanced models are being developed to support robots that can operate safely and effectively in real environments.

3. Change and culture: From adaptation to reinvention

Traditionally, change management focused on helping people adapt to new systems, processes or ways of working, often so the same work could be done more efficiently. With AI, the nature of the work itself shifts. Roles evolve. Accountability changes. Entire parts of the value chain may no longer be human led.

AI capabilities are advancing faster than budgeting cycles and established decision-making processes, culture has to shift with it. Organisations need a culture that can learn, unlearn and adapt quickly. One that’s willing to let go of old assumptions and embrace new ones.

“AI capabilities are advancing faster than budgeting cycles and established decision-making processes, culture has to shift with it.”

This shift also challenges how performance is measured and rewarded. When a junior employee, equipped with AI, outperforms a long-tenured colleague, the old metrics start to fall short. The question becomes: how do we recognise adaptability, skill and contribution in a human and AI workforce?

It’s no surprise that large technology companies such as Microsoft, Google and AWS are leading this shift. What’s instructive isn’t the technology they deploy, but the cultural choices their leaders make: redefining roles, resetting decision rights, and normalising experimentation as part of everyday work. That cultural shift is what allows AI to move beyond innovation teams and into the core of the business.

4. Responsible AI: From guardrails to judgement

As AI becomes more embedded in how organisations operate and compete, responsible use becomes critical to sustaining trust and momentum. Guardrails matter. Strong governance, secure data, and model oversight all remain essential. But responsibility doesn’t stop there. At scale, it increasingly comes down to leadership judgement.

That means asking not only can we use AI, but should we. It’s about using AI intentionally rather than opportunistically, being mindful of environmental impact, and strengthening defences as AI-enabled threats grow. Responsible AI is less about slowing progress, and more about making deliberate, well-judged choices as AI reshapes the business.

Across financial services in Asia Pacific, this is becoming increasingly clear. Institutions like DBS have been vocal about embedding ethical considerations, transparency, and accountability into AI decision-making, recognising that trust is central to scaling AI in regulated environments.⁴ More recently, banks in markets like Malaysia have begun publicly outlining how they are balancing innovation with responsibility, particularly as AI becomes embedded in credit decisions, customer engagement, and risk management.⁵

Here, leadership judgement matters as much as technical controls. Governance helps keep organisations safe. Judgement ensures AI remains aligned to customer expectations and broader societal values.

5. Human-AI teamwork: From assistance to partnership

The real opportunity is shaping work where humans and AI complement each other to achieve better outcomes than either could alone. This is where impact accelerates, and where organisations become more capable, resilient, and scalable over time.

“The real opportunity is shaping work where humans and AI complement each other to achieve better outcomes than either could alone.”

This is especially visible in healthcare. Organisations such as Apollo Hospitals in India are using AI to augment clinicians by supporting diagnosis, triage, and patient engagement, rather than replacing human expertise.⁶ The leadership focus has been on improving outcomes for both patients and clinicians, redesigning workflows so humans and AI operate as partners.

This shift from AI as ‘assistance’ to AI as a true collaborator is where productivity, quality and trust begin to scale together.