Gartner representatives are urging IT leaders in Australia to avoid getting caught up in the tech vendor race to rapidly develop, deploy, and sell AI solutions. Instead, companies should carefully craft either a “steady” or “accelerated” approach to AI adoption — depending on the specific benefits they aim to achieve from the technology.

Speaking at the Gartner IT Symposium/Xpo in Australia on Sept. 9, two distinguished executive analysts at Gartner, Mary Mesaglio and Kristian Steenstrup, explained that tech vendors were going “all out” to develop AI solutions while nearly half of CIOs are struggling to see returns on their AI investments.

The Gartner analysts recommended that Australian organisations focus on their own AI race heading into 2025. But this will require having different approaches to the technology across various IT and business functions.

What is the difference between a steady or accelerated approach to AI?

Gartner’s analysts defined an AI-steady organisation as one that:

• Operates in an industry that is not disrupted by AI.
• Has a modest level of ambition for AI technology.
• Has 10 or fewer active AI initiatives running.

In contrast, those adopting AI at an accelerated pace usually:

• Exist in industries being disrupted by AI.
• Aim to be AI-first organisations.
• Have more than 10 active AI initiatives.

AI benefits: Start with productivity and move to revenue generation

Productivity gains are considered the primary focus for organisations adopting AI at a more steady pace. However, Gartner’s analysts warned the productivity gains from AI are not equally distributed: Most are attributed to employees based on the complexity of their jobs and their level of experience.

“What makes achieving AI productivity easier is matching job complexity and job experience,” Mesaglio explained. “The rule of thumb is you get more AI productivity when you match low complexity and low experience, or high complexity with high experience.”

Companies adopting AI at an accelerated pace are also seeking the same basic productivity gains from AI. However, Steenstrup said this type of organisation is often looking for more from the technology, such as better asset yields, greater speed, new revenue, an enhanced customer experience, and reduced losses.

From proof of concept to proof of value

While organisations adopting AI at a steady pace may only need to scrutinise their AI expenditures more closely, those pursing an accelerated path have been advised to implement real-time cost monitoring — similar to the approach many have taken for tracking cloud expenses.

AI projects also must take cost and value into account from the outset, Mesaglio said.

“When you do a proof of concept, don’t just test whether the technology works and that employees like it,” she explained. “Use the proof of concept to also understand how your costs will scale.”

Build an AI stack to cope with expanding AI tools and data

There has been an explosion of AI features and tools within Enterprise Resource Planning systems, Customer Relationship Management systems, and other external and internal tech tools. This means companies will need to build the capabilities to manage them and organisational data cohesively.

By 2026, Gartner predicts more than 80% of software vendors will embed generative AI capabilities.

Gartner suggests that organisations create a “tech sandwich” to manage both AI and data handled centrally within the organisation, alongside AI and data embedded in other software or “BYOAI” (Bring Your Own AI) brought in independently by different departments in a business.

AI trust: Empower safe and trustworthy AI across the organisation

Steady and accelerated AI adopters must build trust into AI — but in different ways, according to Gartner.

AI-steady organisations

AI-steady organisations can rely more on human-driven governance, policies, and change management to ensure AI safety and reliability. For example, measures such as establishing a responsible AI team for AI safety and creating a community of practice to share expertise can be effective for managing a smaller number of AI initiatives.

AI-accelerated organisations

Gartner argues that those adopting AI more quickly will require a more automated, technology-driven approach to ensuring trustworthy AI beyond just relying on human-governance processes. This means using “trust technologies” that can programmatically enforce AI policies and manage AI risks in real-time.

SEE: Australia proposes mandatory guardrails for AI

AI and employees: Support employee behaviour as AI rolls out

As the Australian government proposes implementing mandatory guardrails for AI, Gartner’s analysts said not enough enterprises consider the emotional impact that the introduction of AI has on employees. They said it could cause them to feel threatened by AI, beholden to the technology, or jealous of their coworkers who are using AI.

“Your change management program is likely not designed to account for the full spectrum of possible emotional responses to AI,” Mesaglio said. “This is way more than UX testing. Only 20% of CIOs say their enterprise is focused on mitigating potential negative impacts on employee wellbeing.”

These feelings may worsen as AI-accelerated organisations roll out agentic AI, or AI agents able to make decisions on behalf of humans.

“We can’t stress enough how important it will be to manage behavioral outcomes with the same rigor as you do technology and business outcomes,” she said.

What aspects of AI must organisations consider in 2025?

Gartner explained to the audience that they do not need to implement AI “all at once.” Mesaglio recommended those planning to implement AI at a steady pace should:

• Seek employee productivity as the main AI benefit.
• Ensure they understand the details of their AI bill.
• Lean on tech vendors to build their AI tech stack instead of building it in-house.
• Continue using AI policies as the main trust mechanism for behavioural outcomes.
• Rely on change management practices — but adapt them for an AI environment.

AI-accelerated organisations should start with the same aims but also:

• Seek benefits beyond productivity, such as improved public outcomes or increased revenue generation.
• Deploy real-time cost monitoring similar to the methods some are using for tracking cloud costs.
• Make a custom AI tech stack that suits the outcomes the enterprise is seeking.
• Introduce trust technologies to automate AI policies and ensure the development of responsible AI;
• Experiment with agentic AI.

Gartner also recommended organisations avoid falling into the yawning trough of disillusionment with AI.

“Everyone talks about the hype at the peak of the hype cycle, but not enough people realise there’s negative hype at the bottom of the hype cycle too,” Steenstrup said. “When you’re in the trough, don’t fall for it. If you stay focused on business value and going at your own pace, you can serve the peaks and the troughs of AI.”