Companies are preparing for agentic and physical AI adoption

• Deloitte finds enterprises scaling agentic and physical AI.
• Productivity gains show, but governance gaps remain.

Deloitte’s AI Institute has released the 2026 edition of its State of AI in the Enterprise report, offering a snapshot of how organisations are using artificial intelligence and what that shift means for leadership, operations, and governance.

Titled The Untapped Edgethe study draws on a survey of more than 3,000 senior leaders involved in AI programmes, including respondents from Singapore. Rather than focusing only on adoption rates, the report looks at how companies are moving from early testing into broader deployment, and what still stands in the way.

According to the findings, many organisations are no longer experimenting at the margins. AI is starting to influence how work gets done. Yet progress is uneven, and leaders face practical trade-offs between running existing systems and investing in new ones.

The report frames Southeast Asia as a region where ambition is already translating into measurable gains. Chris Lewin, AI & Data Capability Leader at Deloitte Asia Pacific, says many businesses are seeing real productivity benefits from early AI investments. At the same time, he warns that scaling those gains requires more than incremental upgrades.

Leaders, he explains, will need “a clear roadmap to guide their transformation beyond incremental optimisation,” supported by governance that is embedded across the organisation. Lewin adds that long-term value depends on rethinking what AI can enable, arguing that “the full value of the technology will come from reimagining what is possible” to achieve sustained competitive advantage.

From pilots to production

One recurring theme in the research is the difficulty of scaling AI projects beyond pilot stages. Early trials are common, but turning them into stable, everyday systems takes planning and discipline.

In Singapore, 32% of surveyed leaders report that at least 40% of their pilot projects have already moved into production, compared with a global average of 25%. Over the next three to six months, more than half of respondents — both locally and worldwide — expect to reach a similar level. This suggests that organisations are under pressure to show results, even as they maintain core operations.

The report describes a risk of “pilot fatigue,” where teams run many experiments without clear direction. Deloitte argues that a structured roadmap can help companies decide which projects deserve long-term investment and which should be retired. Without that clarity, AI efforts can stall despite strong executive interest.

Productivity gains — and deeper change

Many Singapore leaders say AI is already improving day-to-day performance. Seventy-three per cent report gains in efficiency and productivity, above the global figure of 66%. Slightly more than half say AI supports better decision-making by providing data-driven insight.

Yet the study suggests that deeper organisational change remains limited. Only about one-third of leaders say they are redesigning key processes around AI while keeping their business model intact. Fewer are reshaping core operations altogether. This gap highlights a conflict between short-term efficiency gains and broader structural change.

Leaders also point to familiar obstacles. Regulatory and compliance demands top the list, followed by shortages in AI skills and knowledge. Cost and infrastructure limits remain concerns, though they are cited less frequently.

Workforce readiness is another area of focus. Over half of Singapore respondents say building AI fluency is a priority. Many are also reviewing career paths and job structures to account for automationshowing an understanding that AI adoption can change how roles grow over time.

Governance and the rise of agentic AI

The report also tracks growing interest in agentic AI — systems that can act with a degree of autonomy rather than simply offer recommendations. Nearly three-quarters of surveyed organisations expect to deploy such tools in multiple operational areas within two years, far higher than current usage levels.

Customer support, supply chain management, and marketing are seen as likely entry points. Despite this momentum, governance frameworks are still developing. Only a small percentage of Singapore’s leaders report having mature oversight structures for agentic systems. Many organisations rely on a combination of public guidelines and internal policies to manage risk.

Deloitte notes that autonomous agents introduce new governance questions. Companies may need clearer limits on what systems can do independently, along with monitoring tools and audit trails that record decision paths. The aim is to maintain accountability even when software acts without direct human input.

Physical AI enters operations

Another trend covered in the report is the rise of physical AI, which are systems that sense real-world conditions and guide machines or control equipment. Most Singapore respondents anticipate to use such tools within the next two years, indicating a growing interest in automation in industrial or operational settings.

Digital twins, collaborative robotics, and intelligent monitoring are cited as areas where these systems could shape workflows. The report stresses that adoption depends not only on technical performance but also on trust. Secure design, interoperability, and resistance to disruption are seen as basic requirements if these tools are to operate reliably.

Sovereign AI concerns grow

Questions around data control and local infrastructure are becoming more visible in AI planning. A large majority of Singapore organisations say data residency and regional computing capacity matter to their strategy. Many people express concern about reliance on foreign-owned platforms.

Deloitte advises companies to map which workloads must remain within national boundaries and to clarify local compliance rules. Policies governing retraining models, cross-border data movement, and documentation standards may also need review. The goal is to balance operational flexibility with regulatory expectations.

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology eventsclick here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.