At Davos earlier this year, Microsoft CEO Satya Nadella introduced a term that should be ringing alarm bells in every Australian boardroom: “Enterprise Sovereignty.” On the surface, it sounds like standard Silicon Valley jargon. But for a nation of “middle power” industries – our banks, law firms, and miners who often find themselves at the mercy of global software stacks – it touches a nerve, writes David Brudenell.
Nadella’s argument is that companies risk becoming “hollowed out” if they merely rent intelligence from external providers like OpenAI or Anthropic, rather than capturing it themselves. He is right, but for the wrong reasons. The real threat to the Australian economy is about who owns the schema, not the model.
In tech-speak, a schema is structured logic. In the real world of Australian business, it is the “secret sauce.” It’s the decision architecture that determines how a credit officer at a Big Four bank weights a risk, how a partner at a top-tier law firm reads an opposing counsel’s bluff, or how a project manager on a Pilbara mine site overrides a standard protocol based on twenty years of pattern recognition.
This is tacit knowledge. It is the most valuable asset an enterprise owns, yet it is currently being siphoned off in what can only be described as a $1.3 trillion global knowledge heist. The irony is that we are the ones handing over the keys.
Consider a scene increasingly common in Sydney or Melbourne professional services: a senior director needs to map a firm’s expertise for a complex cross-border deal. Ten years ago, that was a day of deep work filtering through mental databases and historical deal sheets. Today, they likely feed those credentials into Claude or ChatGPT and ask it to “format this for a term sheet, weighting infrastructure assets in APAC higher than the Americas.”
In 15 minutes, they have a perfect output. The productivity gain is intoxicating. But in that quarter-hour, the director has also encoded the bank’s institutional logic into a third-party model. Every correction and refinement taught the AI not just what the bank does, but how it thinks.
Once that logic is externalised, it is gone. You cannot put the toothpaste back in the tube. Within 18 months, a mid-tier competitor with zero historical expertise can prompt that same model for a “top-tier deal construct” and get a result that mirrors the elite incumbent. The premium for judgment evaporates, leaving the firm to compete on execution: the very thing AI is currently commoditising to zero.
The scale of this transfer is staggering. If we look at the top 2,000 global companies – many of which have a significant footprint in Australia – the estimated “schema transfer” is roughly $950 million per company, every year. We are trading our long-term competitive moats for short-term OpEx wins.
For Australian industry to survive this shift, we need to stop viewing AI as a productivity tool and start viewing it as a capital allocation decision. Most local enterprises are still stuck in the “data privacy” phase of the conversation. They worry about their servers being in Canberra or Singapore. While important, this misses the point. You can have the most private data silo in the world, but if your employees are using “Trojan Horse” productivity tools to solve their hardest problems, your intellectual property is still leaking into the global commons.
The solution isn’t to ban AI, that ship sailed the moment ChatGPT hit the mainstream. The solution is for Australian firms to capture their tacit knowledge internally before they externalise it.
This requires a fundamental shift in how we manage people and IP. We have spent decades documenting “explicit knowledge”: policies, procedures, and manuals. But the real value lives in the heads of the “indispensables”: those tenured employees in operations or legal whom everyone goes to when things get difficult.
If we don’t build the infrastructure to capture that institutional “know-how” within our own proprietary systems, we will spend the next decade renting our own intelligence back from Silicon Valley at a 10x markup.
The firms that emerge as winners in five years won’t necessarily be the ones with the most “agents” or the biggest tech budgets. They will be the ones that own their decision-making architecture. They will have integrated their human experts at the edges of their own encoded schemas, creating a moat that no generic foundation model can cross.
Renting intelligence is a race to the bottom. Owning your schema is the only way to stay on top.
David Brudenell is Co-CEO of Decidr and Executive Chairman at Decidr AI Industries, where he leads AI enablement initiatives for small and medium enterprises across multiple industries and geographies. With over 20 years of experience spanning technology, data, and commercial strategy, David has held senior leadership roles at Appen (VP, Solutions & Advanced Research), Eclipx Group (Managing Director, Consumer & Chief Digital Officer), Flare HR (Chief Revenue Officer), Universum (Global VP of Product), and Pureprofile (Executive Vice President). He holds an MBA with distinction from Sydney Business School and an Honours degree in Biology from the University of Western Ontario.
Want to see more Forbes articles on your feed? Tap here to make Forbes Australia a preferred source on Google.
Look back on the week that was with hand-picked articles from Australia and around the world. Sign up to the Forbes Australia newsletter here or become a member here.
