Artificial intelligence may be everywhere in today’s headlines, but for governments the real challenge is no longer possibility – it’s execution. In a recent conversation with InnovationAus podcast Commercial Disco, SAP’s Vice President for Global Public Sector Services, Ryan van Leent, explored why moving AI from proof-of-concept to production has become the next critical frontier for public institutions.
The discussion unpacked what it takes to scale AI responsibly, embed it into everyday government operations, and build the trust required for long-term impact: Just because we can doesn’t mean we should’: SAP on AI’s next frontier.
Artificial Intelligence has dominated headlines this year, but for governments and public institutions, the question is no longer what AI can do, but how we make it work in practice and at scale.
At SAP, we recently released our Value of AI Report with Oxford Economics, providing a checkpoint on where organisations stand today. Across Australian government and business, AI now supports about one in four tasks, a figure expected to exceed 40 percent within two years. Encouragingly, three-quarters of organisations anticipate a positive return on AI investment within one to three years.
These findings highlight significant momentum, but also a challenge. Too many AI projects remain trapped in proof-of-concept stage. The real opportunity lies in moving from prototype to production, where innovation can scale, deliver measurable value and build public confidence.
Three priorities for progress
Three priorities will enable Australian government agencies to move AI from experimentation to enterprise-wide impact:
- Adopt embedded AI capabilities. By using applications that already contain AI, public sector organisations can deploy AI easily and scale rapidly.
- Build capability and confidence. Success depends on equipping the public service and its technology partners with the skills to implement, monitor and govern AI responsibly.
- Earn and maintain public trust. AI ethics policies are now widespread, but policy alone is not enough. Trust is earned when we demonstrate that those guidelines and guardrails are being put into practice.
Adopt embedded AI capabilities
Globally, we’re already seeing transformative outcomes being delivered with custom AI.
In Germany, Hamburg’s Ministry of Finance uses machine learning to support processing of social benefit applications, saving 33,000 hours of manual review.
In France, the city of Antibes uses AI to align budgets with the UN Sustainable Development Goals, making over 138,000 decisions with AI assistance, work that would be impossible for humans alone.
But for enterprise AI deployments to become more widespread, organisations need to shift to adopting AI capabilities that are embedded in business applications. For governments, this means switching on the AI that already exists in enterprise applications for HR, finance, procurement and citizen services, rather than building custom AI solutions using technical platforms.
Build capability and confidence
As we move toward agentic AI – systems that operate autonomously – the need for transparency becomes even more important.
We must ensure that AI augments, rather than replaces, human judgment. This means giving users visibility into the reasoning behind AI decisions. At SAP, our applications include AI Analysis that shows which data sources were accessed, what steps the AI took, and how recommendations were generated, giving humans the insight they need to make informed decisions.
Building this kind of transparency into every AI scenario is critical for governments that must demonstrate accountability to citizens.
Earn and maintain public trust
At SAP, we assess every new AI scenario against our global AI Ethics Policy, which is aligned to UNESCO’s Recommendation on the Ethics of Artificial Intelligence.
Several years ago, for example, we developed an “emotional AI” prototype capable of detecting human emotion through facial expression and tone of voice with 70 percent accuracy. Despite its potential, we chose not to productise it, determiningthat the risk of harmful or biased outcomes was too high.
This experience demonstrates how the right kind of regulation can accelerate innovation. Clear guidelines and guardrails focus experimentation on use cases that genuinely improve people’s lives.
Australian Research Alliance for Enterprise AI
SAP is an industry partner in a new Australian Research Alliance for Enterprise AI with the University of Queensland, QUT, UNSW, the University of Sydney, and the University of Melbourne. We’re exploring how AI agents can make enterprise systems more intuitive, responsive and productive across government and industry.
AI’s potential for the public sector is immense, but to realise it, we must move beyond experimentation. By adopting embedded AI, investing in skills, and building public trust, we can shift from isolated prototypes to scalable impact and unlock a new era of productivity and confidence in digital government.
For more information or to engage with the Australian Research Alliance for Enterprise AI: enterpriseai.org.au.
