Companies across Asian supply chains claim end-to-end artificial intelligence, but most real deployment is narrow and targeted, and the gap between the two is organisational rather than technical. That is the assessment of Budi Djojo, group chief technology officer at Ben Line Agencies, the Singapore-based shipping and logistics services group, who has run enterprise digitalisation programmes across third-party logistics, postal, port and maritime operations.

The market noise is loud. Commercial researchers project the market for artificial intelligence in supply chains to grow at 43.9% a year to reach USD 30.81 billion by 2029, according to commercial estimates from The Business Research Company, a London-based market research firm. Vendor marketing promises autonomous, self-optimising supply chains that run themselves.

The adoption data tells a more sober story. In the 2025 Annual Industry Report from MHI, a US material-handling and supply chain industry association, and the consultancy Deloitte, 28% of the more than 700 supply chain leaders surveyed said AI was in use in their operations, while 54% said they planned to adopt it within five years. The report named AI the biggest disruptor of supply chains over the coming decade, yet in the same survey fewer than one in three firms had it in operational use. The distance between what is promised and what is running is the subject Djojo has watched from four vantage points.

A believer in what he calls a “process first, technology second” approach, Djojo told VCA in an exclusive interview that the claims and the reality diverge in a consistent way. “The claims are usually about end-to-end intelligence and autonomous optimisation, while the reality of deployment is much more targeted,” he said. Genuine progress, in his account, concentrates where the problem is well defined, the data is usable and the operational owner is clear. “The gap is not that AI lacks capability; it is that scaling AI across a complex supply chain requires much more than deploying a model.”

The hard part, he argues, is organisational. A successful pilot in one part of a chain does not become an enterprise capability on its own. The difficult work is integration, process redesign, data ownership, decision rights and persuading people to trust and use a new way of working. Supply chains cross functions, companies, systems and physical operations, each with different incentives and levels of digital maturity, so a model that performs in one node can stall everywhere else.

That view makes him cautious about the headline growth figure. He would not equate market growth with deployment at scale. Much of the market, he said, is still in experimentation and early adoption, with many companies running pilots and buying AI-enabled features while they watch for repeatability, implementation complexity and economic value. Fewer have built capabilities that are deeply integrated into operations and deliver measurable value. “The real question is not whether AI works, but where it can create measurable value without introducing disproportionate complexity,” he said. The widest divergence, on his reading, sits around the promise of intelligent, end-to-end autonomous supply chains. “We are not there yet.”

Where AI is genuinely advancing, Djojo points to areas with high operational friction and manageable implementation barriers: document processing, exception management, customer-service support, knowledge retrieval, forecasting and decision support. In each, the workflow is identifiable, the problem is clear and the value can be measured against a baseline. That, rather than the autonomous supply chain, is where the real deployment is concentrated, and it is a useful test of any operator’s claim. A company describing narrow, well-owned automations is further along than one describing a self-running network.

The barriers to getting further are often misdiagnosed. Ranked, Djojo puts clarity of vision and ownership first, then talent, data quality and budget, with culture as the multiplier across all of them. An organisation can have funding, technology and strong partners, he notes, but without a culture of accountability and cross-functional collaboration, AI initiatives stay as isolated pilots. “The stated reasons for digitalisation failure are often budget constraints, poor data or technology limitations, but the actual causes are frequently more fundamental: unclear ownership, conflicting priorities, weak decision rights, and processes that were never properly understood before someone tried to automate or digitise them,” he said. His sharper formulation is that “digitalisation tends to expose organisational ambiguity rather than remove it.”

For an executive evaluating an AI supply chain solution, Djojo’s test starts away from the technology. Begin with the customer outcome, he advises, prove value in the real operating environment, and “scale from evidence rather than excitement.” He would put four questions to any solution: what specific decision, process or customer outcome it will improve; how that improvement will be measured against a clear baseline; whether it can work with the buyer’s real data, systems and workflows rather than only in a controlled demonstration; and who will own the capability after implementation. None of the four, he points out, begins with the technology itself.

The next phase of maturity, in Djojo’s account, will be steadier than the marketing. What separates the companies that create durable value from AI is organisational: clear ownership, disciplined data, and a willingness to change how work is done. Buying more AI-enabled features, he suggests, is the easy part that changes little on its own. A vendor’s demo can prove the technology works; whether the buyer’s own organisation is built to keep value flowing from it, long after the demo ends, is the harder test, and on Djojo’s account it is the one most companies have not yet passed.