Businesses are embracing artificial intelligence (AI), but many are still struggling to translate that momentum into meaningful business outcomes. Only 13 per cent of organisations have reported significant enterprise-wide value from AI, even as most plan to increase investments, according to Accenture.

Accenture’s research found that many organisations that struggle to scale AI tend to deploy AI agents broadly across functions rather than focusing on high-impact decision points – spending more on technology than people and mistaking early pilots for genuine transformation.

SOUTHEAST ASIA’S CHALLENGES TO SCALING AI According to Mr Vivek Luthra, senior managing director, and AI and data lead ( Asia Pacific and Southeast Asia ) at Accenture, organisations that have successfully transformed with AI share five characteristics: a secure digital core, quality data, responsible AI governance, strong CEO backing and skilled teams.

In Southeast Asia, however, building those foundations comes with distinct challenges, starting with talent. “There are not enough specialist AI engineers to go around, and competition for them is intense,” said Mr Luthra. “At the same time, there is a large workforce that is digitally literate but has never worked alongside AI.”Addressing both gaps will require organisations to rethink training and job design, he added.

Southeast Asia’s AI talent shortage and large digitally literate workforce call for new approaches to training and job design, says Mr Vivek Luthra, senior managing director, and AI and data lead (Asia Pacific and Southeast Asia) at Accenture.

The region’s fragmented landscape is another hurdle, with different regulations, data laws, languages and levels of digital development. “A model trained on data from one country may perform poorly in another,” said Mr Luthra. “This is not a problem to work around; it has to be built into the architecture from day one.”Data sovereignty adds another layer of complexity. Some 64 per cent of Southeast Asian organisations plan to increase investment in sovereign AI, keeping data and AI systems under local control and aligned with national priorities.

For organisations operating across the region, this means balancing different regulatory and data requirements while building AI that can work across the business.

Those differences can also influence the type of AI organisations deploy. “In some cases, smaller, specialised models may be a better fit for specific business requirements and could deliver the right outcomes,” said Mr Luthra.

Legacy technology is another major constraint, with many organisations relying on systems that were never designed to support AI.

BUILDING THE RIGHT FOUNDATION Accenture’s Intelligent Digital Brain report identifies four requirements for scaling AI: unified data access, structured institutional knowledge, AI models tailored to specific needs and governance built in from the outset.

For Southeast Asian organisations, putting these foundations in place starts with building a digital backbone that allows data, intelligence and decision-making to flow across the business while complying with regulations in each market. This may mean modernising or retiring legacy infrastructure and moving towards cloud-based architecture. For organisations operating across multiple markets, that architecture must also accommodate different regulatory requirements.

Technology is only part of the equation. Organisations also need to make institutional knowledge – from internal documents and processes to employee expertise – accessible to AI so it can be used consistently.

Building a strong digital foundation involves connecting technology, data and knowledge across everyday business operations.

is designed around this principle. Its AI agents draw on three functions – language, memory and reasoning – to understand an organisation's terminology, retrieve relevant knowledge and use that context to support decision-making.

The Ascott Limited offers one example of how this could work across multiple customer touchpoints. Accenture is working with the Singapore-headquartered global hospitality group to lay the foundation for its next-generation AI-enabled travel ecosystem, connecting core systems across guest interfaces while incorporating responsible AI and data privacy safeguards.

As part of this work, the two companies are piloting model context protocol frameworks and early-stage unified commerce concepts that would allow AI agents to interact with Ascott through large language model chat interfaces, messaging apps and direct booking platforms. The next step is to evolve Cubby, Ascott’s AI travel companion, into a personal travel agent that can compare options, plan itineraries and complete bookings on a guest’s behalf.

THE ROLE OF LEADERSHIP IN SCALING AI Successful AI transformation depends as much on people as it does on technology.

“In too many AI programmes, change management is treated as the last step, but it needs to be the first,” said Mr Luthra. “The speed of adoption will not be determined by how sophisticated the technology is, but by how quickly people feel confident and capable using it.”Equipping employees with AI skills is only part of the challenge. Sustaining change across an organisation also requires leadership from the top. An Accenture survey found that CEO-sponsored reinvention delivered 2.5 times more value than initiatives led at lower levels of the organisation.

“CEO sponsorship unlocks things that technology investment alone cannot,” said Mr Luthra. “These include the permission to challenge how the organisation works, the commitment to invest in foundations that will not show returns in the short term and the signal to the whole organisation that this is not a pilot programme – it is the new direction.”Learn how helps organisations build the foundations for AI at scale.