Critical technologies
ISSUE_05 / AI’s Next Wave: What experts are predicting by 2028
17 June 2026
Jump to:
In the corridors of South Australia’s innovation places, such as Lot Fourteen, the discussion has already moved beyond the AI hype. Experts aren’t just aiming for better chatbots; they are developing autonomous agents and sovereign infrastructure that will reshape the global economy by 2028.
The rise of dynamic agents
One of the newest entrants to South Australia’s innovation places, founded by AI scientists and startup investors, is already looking well beyond the current horizon. LoopSoup founding team members Jo Chaturvedi-Durant, an early-stage investor, and Luke Whitbread, a PhD candidate and researcher at the Australian Institute for Machine Learning (AIML) at Adelaide University, are moving beyond static agent systems to develop what they refer to as a “self-evolving operating layer” for serious enterprise tasks.
The startup’s very name, based on a mathematical theory, hints at the heart of their innovation. The ‘loop’ is the continuously learning AI, while the ‘soup’ is where the dynamic agents operate within a continuous cycle of reasoning, acting, and adapting.
“We are building systems that shift from individual exploration to safe execution,” Whitbread explains, unlike conventional software that follows a fixed, linear script.
What makes these agent systems genuinely dynamic is their ability to rapidly adapt and evolve their environment. In the unpredictable real world, things seldom go according to plan.
A dynamic agent considers a failure a new data point; it can re-evaluate, re-plan its approach, and update its environment and tooling to not only find a solution but also produce increasingly effective solutions that satisfy changing constraints (e.g., governance, safety, or regulatory requirements), without human help. “It’s meant to advance human intelligence and capability, not replace it,” explains Chaturvedi-Durant.
Whitbread describes this as expanding the “cognitive light cone” of what humans can achieve. By allowing AI to handle gruelling processes and tasks, humans are freed to explore territories that were previously “inaccessible” due to time or complexity. This shift supports research co-authored by Chaturvedi-Durant on the “burden of cumulative knowledge”; the idea that humans are increasingly bogged down by the sheer volume of existing data they must remember and process just to reach a new frontier of progress.
The LoopSoup duo says that the future of work isn’t humans talking to AI, but humans managing digital workforces of agents that navigate the unpredictable real world autonomously, unlocking the ability of humans to create greater outcomes.
However, developing this autonomy demands a strong focus on “governance”, going beyond tracing tools to understand why an agent made a certain decision, to systems that continuously evolve to ensure changing requirements are adhered to.
By shifting the burden of storing and processing the world’s data to machines, humans are freed to focus on high-level thinking and new breakthroughs. Value shifts from simply “knowing” to “directing, judgment, and even taste”, with humans setting the overarching intent and guardrails for autonomous agent systems.
The agentic enterpise: Salesforce’s global vision
As local startups innovate, global leaders are shifting their entire models to support this dynamic enterprise. Salesforce, a worldwide corporate brand with a South Australian base at Lot Fourteen, is bridging the infrastructure gap with its Agentforce 360 platform.
Glenn Rozet, Senior Vice President for Australia and New Zealand at Salesforce, predicts that by 2028, isolated “bots” developed in silos will seem as outdated as a paper filing system. Salesforce believes in a future where agents from different departments communicate securely, managing complex citizen requests seamlessly from start to finish without manual hand-offs.
For Rozet, the real worth of government doesn’t lie in the LLM (Large Language Model) itself, but in a trusted, unified data foundation.
By meeting the Australian Government’s national security standards for its platform and agents, Salesforce is giving its public service clients, in particular, the green light to move from experiments to deploying autonomous agents that handle sensitive workflows securely.
“For public service leaders, the goal is to let Agentforce handle the high-volume data crunching so your people can focus on the high-empathy, strategic work that only humans can do,” Rozet says.
Bespoke business and our opposable thumbs
While Salesforce develops the tools, Professor Anton van den Hengel, Chief Scientist and founding director at AIML, foresees a fundamental shift in the very DNA of business by 2028. He highlights a critical turning point where “code” has become a commodity.
“We are at a point where you can write the code yourself cheaper than you can buy it,” Prof van den Hengel explains. He believes the future belongs to bespoke, AI-generated solutions rather than one-size-fits-all platforms.
“Whatever business you name, I can probably recreate its software in a week without any technical debt,” he notes, implying that by 2028, a company’s ability to develop its own unique software will be its greatest asset.
For South Australia, this shift offers a huge economic opportunity. Prof van den Hengel advocates that the state generate and export digital tokens (the units of data that LLMs use to process and generate text) rather than just electricity. By harnessing local wind and sunlight to power large data centres, South Australia can process global AI data and export high-value “processed intelligence”.
“Shipping tokens is much cheaper than shipping electrons,” he notes, positioning the state to be a powerhouse of the next wave.
Electricity for the digital age
Ben Colley, General Manager of Stone & Chalk Adelaide at Lot Fourteen’s Startup Hub, supports emerging technology founders and sees AI as the next major, widely adopted technology, similar to electrification.
“The internet took a long time to spread because we had to build the infrastructure and lay the cables,” Colley explains. “But distributing AI effectively costs nothing. Everyone already has a smartphone.”
By 2028, Colley expects to see “solopreneurs” running multimillion-dollar companies with no employees, supported entirely by a fleet of autonomous AI agents.
For Colley, the next two years are about orchestration, shifting from managing people to overseeing systems of agents. He remains firmly optimistic that the energy of youth at Lot Fourteen, combined with the wisdom of experience, will successfully steer this wave.
In his view, AI is the ultimate tool that will soon be as essential, and as unseen, as the light switch in your office.
A leapfrog moment for South Australia
For Jo Chaturvedi-Durant, South Australia serves as an ideal nursery for this technology. Having experience in Silicon Valley and Singapore, she considers the agile, interconnected ecosystem at Lot Fourteen a unique advantage.
“We have an opportunity to make a significant global impact in this fast-emerging space,” she says, likening the state’s potential to a leapfrog moment where emerging markets bypass a generation of legacy tech to embrace the cutting-edge latest innovation.
