Why APAC Public Sector AI Is Stalling, and What SAP Teams Should Do Next

Public sector AI in Australia is no longer a pilot project. According to new research from Appian, 70% of public sector workers now use AI in their daily work, up from 58% a year earlier. Confidence is climbing too. The survey shows 68% say they understand the AI tools they use in their roles, against 59% a year earlier.

However, the harder truth lies beneath the adoption curve. The same Appian study found that 72% of public sector workers struggle with separate, disconnected databases inside their own organisations, up sharply from 56% in the prior year. Nearly two-thirds (64%) say those fragmented databases have reduced collaboration within their agency, and 53% report working with incomplete or inaccessible information because of data siloes.

Adoption Is Rising, But So Is Fragmentation

This is the paradox APAC leaders need to ponder. Digital investment is paying off in visible ways. For example, 87% of organisations have rolled out new digital initiatives over the past five years, 81% say those projects have improved collaboration, and 86% report that public services are now more accessible to citizens. However, fragmentation is rising in lockstep as new tools are arriving faster than the underlying processes can absorb them.

The study notes that the rise in digital tools alongside growing system fragmentation shows that many public sector organisations are rolling out technology in isolation rather than integrating it in a way that streamlines processes and connects information.

Moreover, AI is falling short when organisations layer new tools on top of disconnected data, legacy systems and manual handoffs, conditions that make it impossible for technology to deliver its full value.

For anyone running an SAP estate, that diagnosis should feel familiar. Decades of bolt-on systems, agency-specific ECC instances, custom Z-code, and shadow databases are exactly the disconnected substrate the study describes.

The broader market tells the same story. A more recent parallel Appian survey of 2,000 US public sector workers found AI shifting from experimental to embedded, with the same three constraints flagged as critical to scaling responsibly:

• Data privacy
• Human oversight
• Legacy system modernisation

Why These Findings Matter

For public sector bodies running SAP, the migration to SAP S/4HANA and SAP Cloud ERP Private (RISE with SAP) is the moment the fragmentation problem either gets solved or entrenched. Therefore, lifting and shifting siloed ECC processes into the cloud without rethinking the workflows relocates the disconnect. SAP Business AI and Joule generate the most value on a clean process layer and a governed data model rather than a patchwork of manual handoffs between agencies and jurisdictions.

Appian encourages organisations to start with their processes, not the technology by identifying where the bottlenecks, delays, and pain points sit. Only then can organisations determine where AI and other new technologies will meaningfully improve the process and deliver lasting impact.

That is a deliberately unglamorous prescription, but it is the right one. In Australian government settings, where scrutiny is high and programs often span multiple agencies and jurisdictions, traceability and auditability are the price of deploying AI.

This is where a process orchestration layer earns its keep. Appian’s model is to connect siloed systems, automate routine tasks and embed AI inside defined, governed workflows rather than alongside them. For a public sector agency operating on SAP, that means stitching SAP S/4HANA, legacy state systems and citizen-facing front ends into one auditable process, so AI acts on connected information with oversight and clear decision thresholds built in from the start.

What This Means for Mastering SAP Insiders

Fix the process before funding the AI model. Fragmentation is rising in step with adoption, so any AI placed on disconnected SAP data risks underdelivering on its promise. The organisation’s Business AI and Joule cases stall if Joule queries siloed, stale records. Therefore, before the next funding cycle, map the processes most critical to service delivery and treat data integration across SAP and non-SAP systems as the precondition for AI.

Treat SAP S/4HANA migration as a process redesign. Lifting and shifting siloed ECC workflows into RISE with SAP carries the fragmentation into the cloud, so organisations pay cloud prices for the same handoffs throttling AI value. Mastering SAP Insiders should use the migration window to consolidate agency-specific instances, retire Z-code that encodes broken processes, and rebuild around clean, end-to-end workflows that AI can safely act on.

Build governance in from the outset, because Australian scrutiny is unforgiving. Auditors will ask how every AI-assisted decision was made, and bolting on transparency afterward is slow, costly, and erodes public trust. Organisations should pair their SAP and orchestration rollout with defined decision thresholds, role-based access and comprehensive audit trails, so AI strengthens accountability rather than introducing uncertainty.