Execution pressures hampering public sector IT modernisation: Report

New research suggests the gap between transformation ambition and execution is widening.

The gap between AI ambition and execution is widening among Australia’s government and critical infrastructure organisations, according to Kinetic IT’s Sovereign Technology Report.

These organisations face mounting pressure to modernise while strengthening cyber resilience operational control without an increase in budget or resources.

However, transformation programs can stall because of operating model issues, skills shortages and unclear accountability, the report noted.

"This is exactly what we see. We talk a lot about driving AI through measurable outcomes, but you can't get to measurable outcomes if accountability is unclear before you start,” said Kishore Jayaram, CTO, Kinetic IT.

"The technology does what it's supposed to do, what fails is the operating model around it: how do decisions get made, who owns them, does the capability exist to sustain what's been built?

“When a program stalls, the honest question isn't what went wrong with the technology. It's who was actually responsible for making it work and whether the organisation was built to support them,” he added.

Five key shifts are shaping Australia’s current operating environment, according to the report, which draws on ADAPT’s 2025 Government Edge research series and interviews with senior technology leaders.

AI adoption is amplifying foundational weaknesses

Agentic AI is a priority, yet only a small percentage of organisations have the governance, data maturity and assurance mechanisms to support safe deployment.

"It's always data but the gap that doesn't get talked about enough is literacy. AI delivers answers with a confidence that gets mistaken for expertise; when that happens, you stop questioning it. You start delegating responsibility to a tool,” said Jayaram.

"The organisations that will scale AI with confidence are the ones building genuine fluency in their people alongside the engineering and platform foundations underneath. Human discernment and judgement remain foundational,” he added.

Execution risk is increasingly limiting transformation scale

Many agencies operate without a formal reinvestment model for maintaining technology assets, despite technology underpinning essential public services and critical infrastructure operations.

Delivery confidence is becoming more important than delivery speed

As platforms, data and AI move closer to the core of how government and critical services are designed, managed, and delivered, public trust is increasingly a key success measure of transformation.

Hybrid operating models are likely to persist for years

There’s likely to be an extended period where legacy systems, cloud environments, operational technologies, and multiple service providers must operate together.

"Hybrid is the reality most agencies are operating in and it's now the steady state. Layering AI on top of that without the right foundations underneath does not equate to transformation. Instead, it creates risk,” said Jayaram.

The role for partners is to help customers maintain coherent visibility and control across the whole environment, not just the parts that are easy to manage.

“That requires forward-deployed capability, deep engineering knowledge, and a genuine understanding of how these environments actually work under pressure. That's the standard we're building to,” he added.

Partner expectations are shifting toward accountability and operational continuity

There are growing concerns around fragmented accountability in multi-vendor environments, particularly during incidents or periods of disruption.

The take-away for partners is that customers want to see trust demonstrated beyond go-live. “It's not enough to deliver. They want to know you'll be accountable for what happens six months later when something doesn't go to plan,” said Jayaram.

"Building that kind of relationship takes consistency, continuity and local people who genuinely know the environment. This matters even more in the AI era because the temptation to delegate responsibility to a tool is real, and our customers need partners who won’t,” he said.