With 42% of Australian enterprises already deploying AI at scale, the next challenge is turning technological capability into measurable operational value. Geoff Thomas, Senior Vice President and General Manager, Infor Asia Pacific & Japan, explains why industry-specific intelligence, governance and process redesign will determine the winners.
Australia’s artificial intelligence journey is entering a more consequential phase. After years of experimentation, proof-of-concept projects and growing investment in AI platforms, a significant proportion of enterprises are now moving towards integrating the technology into everyday business operations. The challenge is no longer simply adopting AI, but making it work reliably, safely and profitably across complex business processes.
The latest Enterprise AI Adoption Index from Infor offers a revealing snapshot of this transition. According to the research, 42% of Australian enterprises have reached full-scale AI deployment, while 85% believe they can manage implementation without disrupting day-to-day operations. Another 82% say their data is sufficiently mature to support reliable AI outcomes.
For Geoff Thomas, Senior Vice President and General Manager, Infor Asia Pacific & Japan, these findings point to a market that is progressing beyond experimentation towards operational maturity. The combination of deployment readiness, data quality and governance is helping Australian businesses translate AI investments into tangible benefits, including average efficiency gains of 34%.
Yet, as organisations prepare to increase spending and explore more autonomous AI capabilities, the next phase will demand a different approach. The focus must shift from deploying more tools to identifying the operational bottlenecks where AI can make the greatest difference.
“AI maturity is about having the data, operational readiness and governance to put the tools you have to work.”
— Geoff Thomas, Senior Vice President and General Manager, Infor Asia Pacific & Japan
Moving beyond experimentation
The distinction between experimenting with AI and deploying it at scale is becoming increasingly important. A successful pilot may demonstrate technical feasibility, but enterprise-wide deployment requires AI to operate within existing workflows, interact with business systems and deliver consistent results without creating operational disruption.
Thomas believes Australian businesses are increasingly demonstrating the confidence and readiness required to make this transition. Rather than treating AI as a collection of isolated applications, organisations are beginning to consider how the technology can become part of the way work gets done.
This shift has implications across procurement, supply chain management, manufacturing and sales. These functions involve substantial volumes of repetitive tasks, information exchange and decision-making. AI can help reduce the time employees spend gathering data, moving between applications and completing routine steps, allowing them to concentrate on activities that require experience and judgement.
However, the opportunity is not uniform across every organisation. Businesses must identify where AI can address a clearly defined operational problem, rather than introducing it simply because the technology is available.
The practical question, therefore, is not how many AI applications an organisation can deploy, but how effectively it can integrate the technology into the processes that underpin its performance.
Finding the right operational chokepoints
As enterprise AI adoption accelerates, organisations risk becoming distracted by the growing number of models, copilots and autonomous agents entering the market. Thomas argues that the next stage of progress will depend on identifying specific operational chokepoints where AI can remove friction and improve outcomes.
In manufacturing and supply chains, these opportunities could emerge across procurement, inventory management, production planning, logistics and order processing. Delays often arise because employees must gather information from multiple systems, verify details or complete repetitive administrative steps before taking action.
AI and automation can help streamline these activities, provided they are deployed with a clear understanding of the underlying workflow.
“Start with the business outcome and then determine where AI and automation can remove that friction.”
— Geoff Thomas, Senior Vice President and General Manager, Infor Asia Pacific & Japan
This outcome-first approach also provides a more disciplined basis for evaluating AI investments. Instead of measuring progress through the number of tools deployed or agents created, enterprises can assess whether a process has become faster, more accurate, less labour-intensive or more responsive to changing business conditions.
The approach is particularly relevant as 68% of Australian enterprises plan to increase their AI investment over the next 12 months. With spending expected to rise, the ability to prioritise high-value use cases will become increasingly important.
Why generic AI falls short
One of the most significant challenges emerging as AI moves into production is the limitation of generic, off-the-shelf solutions. Infor’s research indicates that seven in 10 Australian businesses believe generic AI does not adequately address their industry’s needs.
The problem, according to Thomas, is context.
General-purpose AI can process information and generate useful responses, but it does not automatically understand the specialised terminology, regulatory obligations, operational workflows and data structures that define a particular industry. These limitations become more consequential when AI is expected to execute business processes rather than simply answer questions.
Consider a manufacturing environment where an AI agent is involved in ordering raw materials. For a food manufacturer, the system may need to understand the Brix factor of a sugar shipment, a measure of soluble solids that can be relevant to product quality. In the automotive industry, an agent working with parts may need to interpret a vehicle identification number (VIN) accurately.
These are not interchangeable requirements. Each depends on specialised industry knowledge and the ability to interpret information within its operational context.
The challenge is reflected in Infor’s global research, which found that 73% of manufacturers and 76% of distributors say off-the-shelf AI does not meet their needs.
For enterprises operating in complex environments, this makes industry-specific intelligence an important consideration. AI must understand not just the language of the business, but also the relationships between its data, processes and decisions.
Without this foundation, organisations risk deploying systems that appear capable in controlled demonstrations but struggle when confronted with the complexity of real-world operations.
Trust must be earned through governance
Australia’s AI adoption figures also suggest growing confidence in automation. Only 44% of AI output reportedly requires manual expert review, the lowest proportion among the APJ markets surveyed by Infor.
This could indicate that businesses are becoming more comfortable allowing AI to contribute meaningfully to operational decisions. However, Thomas cautions against interpreting reduced human review as a simple measure of trust.
Reliability, governance and practical experience are closely connected. Businesses cannot sustainably reduce oversight merely by deciding to trust AI more. They need confidence that the technology performs consistently and that appropriate controls are in place when something goes wrong.
Australia appears relatively well positioned in this respect. Only 6% of businesses surveyed reported having no owner for AI governance, risk or compliance.
Nevertheless, governance requirements will become more complex as organisations move from AI systems that recommend actions to agents capable of executing them independently.
Businesses will need to establish clear identities for individual agents, define permissions governing the systems and information they can access, and introduce approval workflows for higher-risk decisions. Comprehensive audit trails will also be necessary to establish what an agent did, when it acted and why.
Executive accountability remains important, too. Infor’s research indicates that globally, only 10% of enterprises have appointed a Chief AI Officer, suggesting that dedicated executive ownership remains relatively uncommon.
The objective is not to insert a human approval checkpoint into every automated activity. Instead, organisations must establish clear boundaries within which AI agents can operate autonomously, while ensuring that people retain oversight where the consequences of a decision demand it.
Cost visibility must translate into business value
Another encouraging finding is that 91% of Australian enterprises say they understand the true cost of AI, placing the country ahead of all other markets surveyed on this measure.
Greater cost visibility allows CIOs and business leaders to make more informed decisions about investment priorities. Yet understanding expenditure is only one part of the equation. Enterprises must also establish whether the benefits justify the costs.
Thomas recommends measuring AI against operational outcomes such as time saved, processing speed, efficiency improvements, error reduction and the amount of manual intervention a process still requires.
The appropriate metrics will vary according to the business function and the problem being addressed. Automating a procurement workflow, for example, may require a different set of performance indicators from improving production planning or accelerating order management.
What matters is the ability to connect expenditure with demonstrable improvements.
“What matters is being able to connect AI investment to a measurable business outcome rather than measuring success by the number of models, tools or agents deployed.”
— Geoff Thomas, Senior Vice President and General Manager, Infor Asia Pacific & Japan
This discipline will become increasingly important as organisations expand their AI portfolios. Additional investment does not automatically translate into better performance, particularly when new tools are introduced without sufficient integration or a clear operational purpose.
For Australian businesses, strong cost visibility offers a useful foundation. The next step is to link that visibility to outcome-based investment decisions.
The emergence of the agentic enterprise
The growing interest in AI agents is bringing another concept into the enterprise technology conversation: the agentic enterprise. However, Thomas believes the term is often misunderstood.
An agentic enterprise is not simply an organisation that deploys large numbers of autonomous agents. If individual agents operate as disconnected solutions, they can create additional technology silos instead of simplifying operations.
The real opportunity lies in enabling people and AI agents to work together across connected business processes. That requires shared business context, integration across the technology ecosystem and consistent governance structures.
It also requires organisations to be selective about where autonomy adds value. Some processes may benefit from agents that can independently complete routine tasks within defined parameters. Others will continue to require human judgement, particularly when decisions involve significant financial, operational or regulatory consequences.
Building an agentic enterprise is therefore as much an organisational and process-design challenge as it is a technology initiative. Businesses must determine how work should be redesigned, what responsibilities agents can assume and how accountability will be maintained across increasingly automated workflows.
“The number of agents an organisation has isn’t a useful measure of maturity. What matters is whether they can work together to make processes faster, more precise and more accountable.”
— Geoff Thomas, Senior Vice President and General Manager, Infor Asia Pacific & Japan
The next three years: Outcomes over technology
Looking ahead, Thomas believes the distinction between AI leaders and laggards will increasingly depend on execution rather than access to technology. As powerful AI models become more widely available, simply adopting them is unlikely to provide a lasting competitive advantage.
The differentiators will be the ability to apply AI to important business processes, provide it with relevant industry and organisational context, and allow greater autonomy without compromising governance or accountability.
Australia’s current position provides a promising starting point. Relatively mature deployment, strong data readiness, widespread governance ownership and improved cost visibility suggest that many enterprises have established important foundations.
But maintaining this advantage will require more than expanding AI budgets. Organisations must connect their investments to measurable outcomes, integrate intelligent capabilities into their operational environments and equip employees to redesign the way work is performed.
For CIOs, this means working closely with business leaders to identify where AI can address persistent inefficiencies and deliver value beyond technology metrics. It also means ensuring that data, integration, skills and governance receive as much attention as models and agents.
The transition towards an agentic enterprise will not be defined by how much AI an organisation owns. It will be defined by how effectively people and intelligent systems work together to improve business performance.
Australia may already be further along the adoption curve than many assume. Its next test will be whether this maturity can translate into a sustained operational advantage. For Geoff Thomas, Senior Vice President and General Manager, Infor Asia Pacific & Japan, the direction is clear: enterprises that focus on the right problems, the right context and the right controls will be better positioned to turn AI investment into business results.