Until recently, corporate sentiment around AI agents was cautiously optimistic. Now, we’re seeing confident action. Our latest international survey shows that 75% of companies have deployed more than one agent—up from 51% just six months ago. 

But across the Asia-Pacific Japan (APJ) region, AI agent adoption is unfolding at two different speeds. Australia is charging ahead, with 88% of executives (compared to 81% globally) trusting AI agents to act on their behalf in a crisis. Plus, 83% of execs say their business would struggle to function without AI agents. Meanwhile, Japan is taking a more measured path. Only 64% of Japanese executives would trust AI agents in a crisis, and 67% see AI as essential to operations, the lowest among all surveyed countries. 

A varying approach across use cases

Across departments, organizations have begun harnessing the power of AI agents to build and act with autonomy. Leaders are exploring new use cases while ramping up existing applications with increased operational maturity.

Coding

Since generative AI entered the scene, software development has been the proving ground for AI adoption. Today, 84% of companies worldwide use AI to write, review, or suggest code. The appeal is clear. Generative coding helps teams move faster, reduce errors, enforce consistent practices, and free engineers to focus on higher value design and problem solving.

In Australia, adoption runs even deeper than the global average: 88% of organizations now use AI for coding tasks. Relatedly, 83% test AI-generated code for accuracy and reliability, but only one third do so through a formal process. This signals that maturity might still lag behind enthusiasm.

In Japan, meanwhile, just 19% of teams conduct consistent, formal code testing for AI-generated code—the lowest among our surveyed markets. 

Incident response

AI’s use in incident response is another marker of growing confidence in this space. AI already helps responders surface logs, past incidents, and dependencies faster. Agentic AI closes the loop between detection and resolution by auto-remediating well-understood issues, triggering failovers, and coordinating workflows across systems.

Nowhere is this evolution more visible than in how executives see the role of AI in crisis scenarios. Globally, 81% of leaders trust AI agents to act on their company’s behalf during an outage or security event. In Australia, that number climbs to 88% of the 250 execs we interviewed. In Japan, meanwhile, just 64% share that trust. 

The pace of AI trust varies across Australia and Japan 

Trust in AI outputs is rising worldwide. Today, 77% of leaders say they’re more confident in AI-generated results than they were a year ago. In Australia, 80% echo that confidence, citing improved output quality and positive hands-on results.

Japanese executives remain more skeptical. Only 69% of executives report higher trust in AI-generated outputs—well below the global average—and just 32% of companies adopted clear guidelines for responsible AI use. 

Ongoing trust building will come with structure. That means stronger oversight, defined testing standards, and transparent approval workflows that give leaders confidence to scale responsibly.

The need for reliability

With AI shifting from experimental playground to everyday infrastructure, executives are increasingly aware of two areas that determine reliability: how well models perform and how consistently tools stay available.

Managing model reliability

Across regions, 85% of respondents believe their organizations need stronger procedures to detect AI errors and failures before they cause disruption. Australian leaders share that urgency with a notable 88% calling for better detection, while 72% of executives in Japan want better detection processes, at a slightly lower level compared to other countries.

Compared to a global rate of 96%, only 89% of Japanese executives say they’re at least somewhat confident their company can detect and mitigate AI failures such as hallucinations or model drift. This signals a gap in operational readiness: Leaders recognize the risks but lack the assurance that existing systems can respond effectively.

Responding to outages and downtime

When it comes to AI tool outages, 85% of Australian leaders report experiencing at least one AI-related outage. That said, 78% of those who haven’t yet faced one already have a response plan in place. That foresight reflects an understanding that reliability depends on preparation.

In Japan, nearly half (49%) of companies have experienced multiple outages, yet only 44% have formal response plans. More than half respond to incidents on an ad-hoc basis, underscoring the need for standardized AI-incident playbooks that treat model and agent failures like any other system outage.

The flywheel effect (and the complexity that follows)

Adoption breeds confidence, which fuels more adoption. Three-quarters of companies now use multiple AI agents, and 25% are running five or more. 

In Australia, the momentum is even greater. A notable 88% of companies have deployed more than one agent, but they are lagging slightly on volume, with only 22% running over five agents. 

Increased adoption also leads to more complexity. Among companies with only one agent, 84% believe AI-driven complexity will outpace available staff. That concern remains high even for those with multiple agents (81%). This shows that even as organizations race to adopt AI, they’re fully aware of the management challenges it brings and are bracing for that complexity to grow.

People and culture keep AI systems resilient

Technology alone can’t deliver reliability. People do. As AI agents handle more of the day-to-day, companies need skilled operators who can monitor systems, validate outputs, and intervene when things go wrong. 

Training staff to oversee agent behavior, interpret recommendations, and fine-tune automation logic will be critical throughout the AI agent adoption journey. So will developing incident and governance teams capable of managing this new digital workforce effectively. 

How does your AI maturity stack up?

AI agents are quickly becoming the foundation of business operations, but across the APJ region, it’s taking shape in different ways. Australia is moving fast, while Japan is still laying the groundwork.

Both approaches point toward reliable, AI-powered operations that strengthen competitiveness and customer trust. Success will depend on balancing innovation with control, building the structures, skills, and partnerships that make adoption sustainable.

Leaders who navigate that balance will set the standard for operational excellence in the APJ region and beyond. Those who wait risk falling behind as AI maturity accelerates around them.

To see what that future can look like, explore the full PagerDuty report.