VET in the Age of AI and Automation

The current wave of technological change is reshaping jobs and skills faster than the system can respond. Here is what to watch for, and how to prepare in 2026.

1. Introduction

We are in a period of rapid change in the way technology is reshaping work, and the demand for skills is shifting with it. AI is already changing the way we work. In the coming 5 years we are going to see AI enabled human avatars become common in the workplace, the evolution of driverless transport, the introduction of humanoid robots, AI agencies augment our work and more capable models that feel like a step toward Artificial General Intelligence. To be honest, it is a little scary and exciting at the same time.

Assuming we avoid a Terminator-style apocalypse, we as a sector need to start thinking about and acting now in order to be ready. As autonomous systems and robotics become cheaper and more capable, employers stop paying people to do the easy and repeatable tasks. The human work that remains is the harder part. So, the role of people is changing, and the standard expected of the person is increasing. This is not a future problem. It is a 2026 problem. This article is a planning brief for the next five years, with a practical horizon of what needs to be in place before 2027. The aim is not to speculate about far-off futures. It is to help RTO leaders and people considering opportunities in the sector make better decisions using the forces already visible now.

Two pressures are arriving at the same time. Australia’s workforce supply is tightening as retirements rise, skilled migration is becoming more constrained politically and work participation patterns change. At the same time, AI and automation are lifting productivity and reshaping work. This does not simply remove jobs. It changes tasks, compresses entry-level work, and increases the value of people who can supervise systems, handle exceptions, and monitor quality and safety. That combination will change the training market.

If you lead an RTO, or you are considering investing time or capital in VET, you have a choice. You can wait and respond to this demand shift once it shows up in enrolment numbers. Or, you can adapt your market offering now, while you still have room to make deliberate decisions about your market offering and employer partnerships. The question this article answers is not “Will technology disrupt jobs?” The better question is “How will demand for VET change in the next five years, and what should we build before 2027 to stay relevant?” Each section links a real-world driver to a demand shift, then turns that into practical implications for programs, learner cohorts, and the operating model required to deliver credible competence at pace.

2. Why demand for VET is changing

VET is not a single market. It is a set of markets that behave differently as the labour market tightens or loosens. If you have been in the sector for any length of time, you have seen this pattern. When unemployment is high, training linked to jobseekers lifts and is supported by government funding to maintain revenue from income tax. Conversely, when unemployment is low, that funded volume tends to fall because fewer people need training to re-enter work. Government support usually reduces and industry play a bigger role in funding their own training.

This cycle of demand matters because many RTO business models are built on that cycle. Some organisations are heavily exposed to jobseeker cohorts and employment-services linked activity. Others are mainly supported by employer funded training, regulatory and licensing requirements, or employer partnerships with planned intakes. These RTOs experience the swings in employment very differently and both experience peaks and troughs. But things may be changing due to economic shifts in demographics, work fragmentation and technology driven automation.

It helps to think about unemployment and labour availability as the result of three forces working together.

Workforce supply is shaped by retirements, participation, skilled migration and the size of the working-age population. When more people exit work than enter it, replacement demand stays high and labour markets remain tight even when growth slows.
Work fragmentation. More people are working across multiple roles, contracting, entrepreneurship, or using flexible arrangements. This does not always reduce headcount, but it often reduces dependable hours, stability, and willingness to take on supervision and training burdens.
AI, workflow automation, and robotics lift productivity. In some areas that reduces labour demand. In other areas it changes job design, reduces routine tasks, and increases demand for higher-skilled oversight and exception handling.

The key takeaway for providers is this. The employment cycle that we have known and relied on for the past three decades is changing. We have the dynamic of shrinking working population, reducing dependable hours of work per person and massive changes in the nature of work caused by technology. This means that the old demand pattern is becoming less reliable. The practical result is that demand shifts toward employers retraining and retaining the workforce they already have, not just training new entrants or people between jobs. Training demand also becomes shorter-cycle and more frequent, with greater emphasis on step-up capability, cross-skilling, and the skills needed to work safely and effectively in more automated workplaces. We are also likely to see a greater demand for non-accredited training as the national skills framework fails to keep pace with the rapid changes brought on by the waves of technological change that are coming and in some cases are already here.

3. How demographic factors are influencing demand for VET

To understand what is changing in training demand, it helps to consider the demographic settings that sit underneath the labour market. Treasury’s 2023 Intergenerational Report (IGR) describes a population that is ageing and growing more slowly, mainly because Australians are having fewer children and living longer. Between 2022-2063, the IGR projects the share of Australians aged 65 and over will rise from 17.2% to 23.4%, while the share aged 15-64 falls from 64.7% to 61.2%. The point is not the long-range endpoint. The point is that the working age share is shrinking while the older share grows, and that changes the labour market conditions that sit underneath VET demand.

The same report also shows why “workforce tightening” is not just about headcount. Even though participation has been strong in recent years, the IGR projects the overall participation rate gradually declining from 66.6% in 2022–23 to 63.8% by 2062-63, mainly because a larger share of the population is older and participation drops as people move toward retirement. It also notes that employed people are working fewer hours on average, and projects average hours worked per employed person edging down from around 32 to around 31 hours per week by 2062–63. In plain terms, Australia can have plenty of people “in work”, while still having less reliable coverage and less experienced capacity available on the ground.

Fertility is the other structural pressure. The IGR notes that Australia’s total fertility rate has been below the replacement rate of 2.1 since the 1970s, and it assumes fertility will decline from 1.66 babies per woman in 2022-23 before stabilising at 1.62 from 2030-31. Lower fertility matters because it reduces the long-run flow of new entrants who eventually replace retiring workers. Over time, that increases “replacement demand” even in periods when overall job growth is weak.

Migration can ease the squeeze, but Treasury is explicit that it is policy and cycle dependent. The IGR expects a temporary catch-up after the pandemic and says net overseas migration is expected to “largely return to normal patterns from 2024-25”. It then assumes net overseas migration falls back to 235,000 per year in the long run, while noting outcomes depend on future migration policy settings and economic conditions. It also seems politically popular to tighten migration settings further, which may continue. The practical planning point for RTO leaders is that migration is not guaranteed to prop up demand like it has in the past. It helps, but it is not something employers or providers can bank on to solve skill and experience gaps quickly.

Put together, these settings create a very specific labour market pattern over the next decade. Employers will spend a lot of their effort replacing those retiring and lifting capability in the workforce they already have, because experience and supervision become harder to find. For VET, that is the heart of the demand story. More training for existing workers stepping up, more cross-skilling and conversion, and more focus on verified competence where experience and judgement are scarce.

What all this means for the labour market:

Backfilling vacancies becomes the main hiring task. Even if the economy slows, employers still have to replace people who retire or leave. A large part of recruitment is not about growth. It is just about keeping the lights on.
Skill shortages hit hardest where you need experienced people on the floor. If a role needs supervision, a licence, or safety sign-off, you cannot replace experience quickly. When senior people exit, the gap is not just numbers. It is judgement and responsibility. Difficult to replace.
Good people get harder to find, including trainers and assessors. When the labour market is tight, everyone is competing for the same experienced staff. RTOs feel that directly because delivery depends on those people.

What this means for VET demand:

More training for people who are already working. Organisations will train existing workers to step up, because recruiting supervisors and leading hands is hard.
More demand for “conversion” training. Employers will move people sideways and up faster. They will want recognition and short gap training that gets someone competent in a related role without starting from scratch.
More demand in regulated and high-risk sectors. Where safety and duty of care matter, employers cannot afford skill gaps. That keeps demand strong in care, community services, construction, logistics, and utilities.

The practical implication for RTOs is this. Employers will value training that produces a reliable flow of competent people into real roles. That means partnership programs with employers, training that is internally integrated, and strong completion support. Broad scope and generic offers matter less when employers want a training partner that can build capability fast and reduce churn. There will be less demand for qualifications and more demand for shorter training programs that are customised for the employer’s needs.

4. The fragmentation of work

Work is changing not only because of technology, but because of how people choose to participate in work. In many industries, more workers are splitting their time across multiple jobs, doing contract work, or building small side businesses. Remote and hybrid work has also stuck for roles that allow it post pandemic. The result is not that people leave the workforce. The result is that fewer people are tied to one employer with stable, full-time hours for long periods.

This shows up in the labour statistics with part-time work now sitting at about 31.2% of employment (trend, December 2025). Multiple job holding is around 6.5% (September 2025). Casual employment is still a large slice of the workforce. Over the longer term the bigger structural shift is away from standard, permanent full-time work. More Australians are spreading their work across multiple arrangements, and fewer are anchored to one employer on a stable roster for years at a time.

For employers, this is a capacity and reliability problem. It is harder to secure dependable hours, it is harder to build stable teams, and it is harder to find people willing to take on supervision and responsibility. Even when unemployment is low, employers can still feel short staffed because they cannot get coverage and continuity. This matters because it changes how employers think about training. When retention is uncertain, employers become less willing to fund long, broad training that takes months to pay back. They want faster capability, tighter role fit, and proof that the training will translate into performance quickly.

For VET demand, fragmentation pushes the market toward a different shape. You see more demand for short, targeted upskilling that helps people move between tasks, sites and clients. You see more demand for recognition and gap training, because workers and employers do not want to start again every time a role changes. You also see more demand for practical business capability that sits alongside technical skill, because more people are working semi-independently. That includes client handling, quoting and scheduling, basic documentation, and role-specific compliance.

The bigger demand shift is who pays for the training. In a fragmented workforce, demand shifts away from one employer sponsoring a long training program, and toward shorter purchases made close to the point of need. Sometimes the individual pays. Sometimes an employer pays for a specific gap because they need the capability now. Sometimes an industry body funds training as part of some industry initiative. For RTOs, this changes the commercial competition of the market. The winners will package capability in smaller blocks that are quick to deliver and clearly linked to work outcomes, and they will make those skills easy for the next employer to recognise.

5. Three technology waves that are changing the nature of work

If you want a practical way to think about how technology is going to impact over the next 5 years, focus on adoption speed. Some changes spread quickly because they are software based and can be rolled out through existing services. Others move slower because they rely on hardware, capital investment, safety controls, and regulation. That difference matters for RTOs because it tells you where training demand will shift first.

There are three technology waves of change worth watching.

Wave 1 is agentic AI and autonomous workflows. This is the fast mover because it is software deployed and is already happening. Agentic AI links tasks across systems and runs routine workflows end to end, not just one step at a time. In practical terms, it can triage a request, pull the right information, update records, trigger the next action, and escalate when it hits a rule boundary. It spreads quickly because organisations can add it to the systems they already use, without buying new physical equipment. The key shift is that people do less routine coordination and more work setting the rules, monitoring performance, and stepping in when only when required.
Wave 2 is conversational avatars and digital humans. These are human-like interfaces for routine, policy-based interactions, such as reception, triage, bookings, directions, and standard enquiries. This is like talking to Siri in a digital human like form, only 100 times smarter and more capable than Siri. Digital humans are being adopted already in controlled environments because the organisation can tightly manage what the avatar can say, what it can access, and when it must hand off to a person. Increasingly, these avatars are paired with agentic AI, so they do not only answer questions. They can also log requests, update systems, and complete routine transactions within set rules. The human role shifts from delivering predictable frontline interactions to managing exceptions, and to setting up, testing, and monitoring the avatar system.
Wave 3 is physical automation, including robotics and autonomous vehicles. This wave arrives later than the first two because it needs more than software. It needs hardware, capital investment, safety controls, site redesign, and regulation. Adoption starts where environments and routes can be tightly controlled, then spreads as costs fall and reliability improves. It will spread fastest in controlled environments and defined industries, such as warehouses, logistics hubs, depots, ports, mines, and freight corridors. It will move slower in busy, unpredictable environments where safety risks are higher and there are more unusual situations to handle. Autonomous vehicles are already in use in Australia within controlled settings such as resource extraction. There are already training products and site-based training practices in mining that support autonomous operations and we are moving toward defined-route operations, and 2027 is a key policy milestone. That makes autonomy a near-term planning issue, not a distant one.

Across all three waves, the near-term effect is not mass job extinction, it is job redesign. Routine tasks get automated first. Human work shifts to what is still hard to automate, setting up the AI workflow, monitoring what the system does, and stepping in when quality or safety is at risk. This is the demand shift that matters for VET in the near term. Employers will increasingly need workers who can operate alongside automated systems.

The training market will need to make this transition with industry. This requires a shift away from teaching people how to do routine tasks, and toward teaching people how to work alongside automation, supervise AI systems, handle exceptions, and monitor the performance of automation. That is a big paradigm shift that most providers have yet to really grasp and the early adopters will be the ones that benefit the most.

6. Agentic AI and its integration with work practices

Most people now understand generative AI as a tool that creates content. You may have already seen how this has changed the performance of work. Generative AI learns patterns from large amounts of existing data (text, images, audio, code, video). Then, when you give it a prompt, it produces an output that matches those patterns. Common outputs include written text, summaries, images, music, code, and synthetic data. It predicts what should come next based on the input and what it learned during training. That makes generative AI useful for drafting, ideation, translation, tutoring, and automation support.

Agentic AI is different. Agentic AI builds on generative AI with automation rules and boundaries. It automates chains of work across systems, not just single tasks. An AI agent can be given a goal, work out the steps, pull information from the right places, update records, trigger communications, and escalate when it hits a rule boundary. In plain terms, it turns a workflow into something software can run, with humans supervising it.

Agentic AI is happening now because we already have all of the infrastructure in place to support its deployment. Most common applications and software we rely on are cloud based, most support Application Programming Interface (API) which allows an agent to communicate and transact with these applications. You can see it already in service desks, rostering, finance operations, compliance workflows, service scheduling, basic procurement, early-stage recruitment coordination, sales and marketing operations, enrolment processing, and general administration. Once organisations start using agents or teams of agents, the routine parts of these workflows shrink quickly. A receptionist that was performing these tasks moves to dealing with more complex tasks and problem solving.

People spend less time on triaging, follow-up, chasing information, updating systems, and basic documentation. The work shifts to setting up the workflow, deciding on the rules, and controlling risk. Organisations start caring about permissions, approvals, guardrails, logging, and monitoring because they need confidence that the agent is acting within policy. They also need confidence that someone can step in quickly when the system makes a mistake, or when a situation does not fit the standard pattern.

Even though this is happening right now, we are not prepared. Think of the units of competency you are delivering in business, finance, information technology, etc. These training products read like it is 2010. They do not prepare people to work alongside these AI agentic capabilities. I need to be clear here and make the point, working with agentic AI is not a specialist skill, it is a core skill and right now, our national training framework barely recognises it. We are way behind where we should be and therefore; the early adopters are making it up as they go. Training organisations may be asked to customise nationally recognised training to cater for agentic workflow. This becomes tricky because the RTO needs to deliver the unit of competency and so the solution at the moment looks more like non-accredited training.

AI literacy needs to be a core skills across business roles because people need to understand what these systems can and cannot do, and how to use them safely. Demand will grow for agentic workflow and process skills, because someone has to map the work, remove choke points, set handover points between humans and automation, and define what “good” looks like. There is also more demand for applied judgement, because as routine tasks are automated, the human work becomes verification, escalation, and decisions that affect customers, safety, privacy, and cost. Judgement and decision making when the agent hits a rule boundary is going to become the critical skill that needs to be trained through repetitive scenario based training. For RTOs, the opportunity is to train the people who will design, implement, and supervise these workflows and to provide gap training to existing workers to augment their current skills. That capability will sit with experienced workers and with junior to middle management, because they are closest to how work is actually done. Training does not need to be technical to be valuable. It needs to match workplace problems.

In the Building an AI-enabled workforce: impacts for Finance, Technology and Business education and training (Future Skills Organisation, Feb 2025), the report states 86% of workers are interested in receiving more AI training, and also that half of workers report they have not been provided with any training or resources to understand and use AI tools. There is demand now that needs to be met. Agentic AI will not only change what work looks like, it will change what training employers buy.

7. Digital humans and conversational avatars

Conversational avatars and digital humans bring a human-like interface to routine, policy-based interactions. Unlike a chatbot that sits as a text box on a website, a digital human avatar uses voice and real-time digital interaction and can be deployed in places where people normally expect to deal with a person. That includes reception, customer service triage, bookings, directions, and student support, including after-hours and 24/7 study support. The AI can be an expert on whatever topic or learning area you can think of. At the end of the day, what knowledge it does not already hold in its language model, you can provide it in a knowledge base. Imagine the student having a digital human trainer that the student can access anytime and ask questions specifically on your course content?

The combination of a digital human with a deep language model and agentic AI matters because they replace the predictable part of frontline work. They will be adopted first where the environment and the rules can be controlled. Reception and visitor management in large sites is an obvious starting point. Kiosk-style service points are another. Internal student support that handles common enquiries are also well suited. In each case the organisation can tightly manage what the avatar can say, what it can access, and when it must hand off to a human. The more recent shift is that digital humans are not only “a talking head”. In many implementations they can sit on top of agentic workflows. Within defined permissions, they can look up information, log requests, take an order, update records, accept a request, trigger a process, and escalate when a request sits outside the rules. That expands their impact from answering questions to completing routine transactions.

I predict that even regulators like ASQA will introduce digital human avatars to perform preliminary frontline work of interacting with RTOs on enquiries and routine administrative tasks. Your opening meeting and request for evidence could easily be handled by a digital human. These models are already capable of complex lifelike conversations and their knowledge base can be customised specifically for any organisation or industry. Check out platforms like Unith AI or UneeQ. You can already see how the demand for skills will change. When the predictable interactions are handled by a digital human, the remaining human work becomes exceptions, complaints handling, supporting vulnerable customers, and situations where the organisation carries risk.

A second demand area sits behind the scenes. Organisations need people who can set these systems up and keep them reliable and secure. Again, this is not an IT task but a core skill for experienced workers and junior to middle management. The work moves toward writing and maintaining the knowledge base, defining rules and escalation pathways, testing, deploying, monitoring performance, and fixing failure events. Over time, it is reasonable to expect large service organisations to increasingly use digital humans for high-volume, routine enquiries and administrative interactions, while humans focus on complex cases and decision making. Over time, the cost of these services will come down and we will see them creep into medium and small sized organisations.

The demand for frontline service training will gradually decrease and evolve. RTOs that rely on these markets will have to adapt to survive. You should be actively investing in developing new training pathways that the market will be looking for. You will make mistakes and need to test and adjust but, the early movers will be the winners. This may require non-accredited training as an offering in the early stages in order to get a presence in the market. Over time this may convert to a nationally recognised training but, I would not be sitting around waiting for Jobs and Skills Councils to catch-up. At a minimum right now, you need to be investigating these developments and monitoring their development and adoption so you can remain aware and be better positioned to see the opportunities as they arise.

8. Physical automation with robotics and autonomous vehicles

Physical automation is where change becomes visible. You can see the machines. You can see the workflow change. It also becomes obvious why adoption is uneven. Robotics and autonomous vehicles work best where the environment is controlled, tasks are repeatable, and safety risk can be managed through clear rules. That is why change tends to arrive first in warehouses, logistics hubs, depots, ports, mines, manufacturing, food production and freight corridors.

Trade related work often sits at the other end of the spectrum. Plumbing, bricklaying, carpentry, electrical work on building sites is messy and variable. The job changes hour to hour. The space is irregular. The work is tactile. Robotics will still enter these industries, but it tends to land first on the repeatable parts of the job and in controlled settings. The complex work, exceptions, and responsibility for quality and safety stays with people for longer.

It helps to understand how robots learn, because it shows which parts of the job robots can pick up quickly, and which parts will still need skilled people. Many systems learn from observing demonstrations and written instructions. A robot can be trained to learn a task from demonstrations and practice, and it can improve with feedback, but the harder the task and the messier the environment, the more data and control it needs. The limitation is variation. Contact-rich work needs sensing, calibration, and a lot of training data to handle different conditions reliably. That is why near-term gains are strongest in predictable tasks such as moving goods, sorting, pick-pack, palletising, cleaning in controlled spaces, fast food or packaged food production and depot to depot movement.

Autonomous vehicles sit in the same category. In the near term, they roll out fastest where routes are defined and risk is manageable, such as depot-to-depot freight, freight corridors, controlled sites, and some structured last-mile operations. In November 2025, Transport Ministers agreed to allow conditional deployment of automated vehicles from 2027 in selected locations, subject to states and territories updating their legislation and capability. In the United States, driverless taxi services are already operating and expanding, which is a useful signal that this is moving from trial to everyday use in some markets.

It is quite plausible that the occupation of a professional driver will largely become redundant in the next decade. If you train people for work in predictable environments, you should have your eyes wide open. Roles most exposed are those built around repeatable manual handling and rule-based work, for example:

driving in depot-to-depot freight movement
warehouse operators doing pick-pack, and sorting
factory line workers and machine-tending roles
commercial cleaning in controlled sites
laundry processing roles in commercial laundry facilities
product packing and palletising roles
retail stock replenishment and back-of-house handling
fast food or packaged food production roles
security gatehouse and access control roles in controlled sites
traffic control roles in controlled sites

The point is not that these industries disappear. The point is that the task mix shifts. Robotics and autonomous systems do more of the routine work, and people shift toward setup, supervision, fault finding, safety, incident response, and keeping operations reliable. That creates a clear training demand pattern. Demand grows in automation safety, human-machine work practices, diagnostics, maintenance, and supervision. Training in these at-risk roles does not vanish, it moves up the skill curve, and becomes more about systems, oversight, and response rather than repetition.

You should be actively investigating your own industry to become aware of how robotics and automation is impacting your own industry and the training you deliver. These are some questions you should be asking:

Which tasks in our industry are most repetitive and predictable?
Where do those tasks happen in controlled conditions?
Who are the businesses developing these technologies and what are they claiming and predicting themselves?
What would change in roles if tasks were automated?
What new skills would workers need to work alongside these technologies?
What does this mean for training demand in the next 2–3 years?

This space is moving so quickly that you need to be re-evaluating every six months. Subscribe to trusted reliable information sources in your industry and in the tech space so you can stay on top of what is happening.

9. The workforce you already have is the workforce you will be relying on

It would be naive to ignore that some jobs will be displaced or disappear as AI and automation spread, even though new roles will also be created. The World Economic Forum’s Future of Jobs Report 2025 estimates that “job creation and destruction due to structural labour-market transformation will amount to 22% of today’s total jobs”. So almost a quarter of the global workforce is going to be reshaped through either job creation or elimination. It also projects that “170 million jobs are projected to be created and 92 million jobs to be displaced” which implies a net increase in jobs overall, alongside significant churn. In the same report, 41% of employers say they plan to “downsize their workforce where AI can replicate people’s work”, while 77% plan to “reskill and upskill” their workforce to work alongside AI (Fig 4.14).

The International Labour Organization’s Generative AI and jobs: A 2025 update (2025) takes a fairly measured view of how the workforce will be impacted. It finds that “one in four workers across the world are in an occupation with some degree of GenAI exposure”, but it also states that “because of the continued need for human input, most jobs will be transformed rather than made redundant”. Exposure is not evenly spread. The report indicates higher exposure in high-income countries, reporting a 34% share of employment in exposed occupations in those economies, which is directly relevant to Australia.

McKinsey’s modelling in Generative AI and the future of work in America (2023) is based on the United States, but it is a useful window for Australia because technology adoption and operating models often flow here with a lag. McKinsey estimates that “by 2030, activities that account for up to 30 percent of hours currently worked could be automated”. It also expects major labour market movement, stating “we expect an additional 12 million occupational shifts by 2030”. In other words, around 12 million US workers would need to move into different types of roles as demand falls in some jobs and rises in others. That is roughly 7.1% of the workforce changing occupation. If you apply the same proportion to Australia, it is about 1.1 million people needing to move into a different type of role by 2030. The report also flags where disruption is likely to concentrate, stating that “the biggest future job losses are likely to occur in office support, customer service, and food services”.

AI and automation will change employment in different ways. Some jobs will shrink or disappear. Many jobs will stay, but the work inside them will change. Most employers will see this first as a shift in tasks. Systems take over the routine steps, and people are left doing more oversight, checking, and dealing with the tricky cases.

This is landing at the same time as Australia’s workforce is tightening. More experienced workers are moving toward retirement, and it is harder to replace capability quickly. Employers can sometimes recruit a person, but they cannot recruit experience on demand. That is why retention matters more than it used to. Losing a capable worker is not just a vacancy. It is lost judgement, lost supervision, and lost capacity to coach others.

Put those pressures together and you can see where training demand will go next. Employers will spend less time looking for perfect recruits and more time building capability in the workforce they already have. They will buy short, practical training that helps staff get value from new technology in day-to-day work, follow new processes, and make good calls when the system does not deliver the right result. That demand will be strongest for supervisors and team leaders, because they are the ones who keep work stable when roles change and teams churn. For RTOs, this is the signal to follow. Build offerings for employed workers, keep them tight and job-based, and link them to real workplace problems as technology is introduced.

10. What training will be in demand

Before we move into the final sections, I want to recap a few key points. The employment cycle alone no longer explains what drives demand for training. Demographics are tightening workforce supply, work is becoming more fragmented, and automation is changing the task mix inside jobs. We have also stepped through the three waves of technology that will land at different speeds, from agentic workflows to digital humans, to physical automation. The common thread is that employers will not be able to recruit their way through this change. Most demand will come from retraining and lifting capability in the workforce they already have. With that context in mind, the question for RTO leaders is practical. What should you be watching for next, and how will demand shift as these changes move from early rollout into everyday work?

Training demand is already shifting in ways that will show up in enrolments, employer enquiries, and funding opportunities. The change is not evenly spread. It will hit some roles and some industries first, then widen. If you want to stay ahead of it, watch what employers are asking for, not what the technology marketing says.

First, expect demand to move up the workforce. The early lift is not from new entrants. It is from employers trying to keep their current workforce effective while tools and processes change. Employers will look for short, practical training that helps staff use new systems well, follow new workflows, and deal with the tricky cases that the system does not handle cleanly. This is strongest in business functions that are heavy on routine information work, for example administration, customer support, finance operations, rostering, scheduling, and basic compliance processing.
Second, expect a squeeze in entry-level work in office roles. Many entry-level jobs in clerical and support functions have been built around routine tasks. As agentic workflows and digital service tools absorb more of that work, the entry-level role changes. It becomes smaller, or it becomes more demanding earlier. Employers will start asking for new starters who can handle more responsibility sooner, which pushes demand toward structured pre-employment preparation and tighter “job-ready” capability in basic business systems, communication, and problem handling to work alongside AI capabilities.
Third, expect “conversion” training to become a steady market. As tasks shift inside roles, employers will move people sideways and up more often. They will want recognition of existing capability and short gap training to get someone competent in a related role without starting from scratch. This will show up as increased demand for short, targeted programs that bridge between related jobs, for example from admin to customer operations, from customer support to service coordination, or from basic finance processing into higher-value checking and exception handling. Again, the integration of AI literacy into these programs is going to be critical.
Fourth, expect shorter training products to outsell full qualifications in the early stages. Many employers will still want nationally recognised outcomes where they are required. But when the need is “we have changed our workflow this quarter” they will not wait for a long program. They will buy shorter training that is specific to their systems and processes, and they will want it delivered quickly. This is where you will see more demand for customised delivery, short skill sets, and add-on capability that sits alongside nationally recognised training. You may need to customise AI workflow training alongside nationally recognised training programs.
Fifth, plan for non-accredited pathways as a bridge while the national system catches up. Jobs and Skills Councils and the training product development cycle are not built for monthly or quarterly change. In fast-moving areas, there will be a gap between what employers need now and what can be turned into national training products. Employers will want training immediately, because their systems and workflows are changing now. In many cases, there will not be a nationally recognised training product that matches what they need yet. That creates a gap that RTOs can either ignore or fill. The providers who respond will be those who can design and deliver fit-for-purpose non-accredited training quickly, explain it clearly to employers, and, where it makes sense, bolt it onto existing programs as AI literacy and safe-use training. This is also the area to watch for policy change. Industry is likely to push for funding that supports targeted non-accredited training when the need is urgent and clearly linked to workforce productivity.

If you want a simple set of indicators to track in 2026, watch for these shifts in your employer enquiries:

Requests for short, role-specific upskilling instead of full qualifications
Increased focus on supervisors, team leaders, and coordination roles
More demand for recognition and short gap training to redeploy staff
More questions about AI tool use, workflow changes, and how to supervise automated processes
More interest in non-accredited training or add-on skill modules that can be delivered fast

This is where demand is heading. It is practical and it favours RTOs that can respond quickly without losing credibility.

11. What to do in 2026 to be ready for 2027

Technology-led change is accelerating, and it will keep widening in scope over the next few years. It will reshape what employers ask for and how they buy training. If you want to be ready for 2027, treat 2026 as a build year and take a small number of practical steps now.

1 – Decide what you will be known for. Pick 3–5 programs or capability areas you will specialise in and start investigating how AI is going to disrupt those capability areas.
2 – Build a non-accredited offer you can deliver fast. Create a small set of short, non-accredited modules you can bolt onto your programs or deliver to employers now, with a focus on AI literacy, and working with automated workflows.
3 – Strengthen your “job-based” delivery model for employed learners. Make it easy for employers to buy short training for existing workers. Offer short blocks, flexible scheduling, and clear outcomes that map to real work tasks.
4 – Lift supervisor and step-up pathways. Build a simple pathway for step-up training (leading hand, team leader, supervisor), with scenario activities focused on applying judgement, exceptions, and oversight of AI workflows.
5 – Set up two employer partnerships with planned intakes. Secure two employers where you can support ongoing intakes and workplace support, not one-off enrolments. Keep it simple: agreed numbers, timing, and support arrangements. These partnerships are going to be critical in the future.
6 – Use AI in your own operation, properly. If your team is not already using AI, start now. Put simple rules in place for safe use, then use generative AI to reduce admin load and improve consistency. Next, pick one or two workflows where you can trial basic agentic automation, such as enquiry triage, enrolment processing, scheduling, or internal support. The goal is to build real capability and confidence inside your RTO, not just talk about it.

If you do these six things in 2026, you will enter 2027 with a sharper offer, faster response capability, and stronger alignment to the demand shift already underway.

12. Conclusion

We are on the cusp of a level of change that most of the training market has not dealt with before. AI is already reshaping office work. Digital humans are starting to move into service environments. Physical automation is moving from pilots into real operations. The pace will increase, and the scope will widen.

For RTOs, the risk is not that training disappears. The risk is that demand shifts and you do not notice until your numbers start slipping. Employer enquiries change first. They stop asking for long programs and start asking for short, practical capability that fits new workflows. They start asking for help retraining the workforce they already have. They start asking for fast solutions that the national system cannot package yet. If you keep offering what you have always offered, revenue does not collapse overnight. It quietly leaks away to providers who adapt faster.

This is the moment to have your eyes wide open. Track what is happening using credible sources. Talk to employers about how work is changing in their operations. Test new offers in small blocks. Build confidence inside your own business by using AI properly. Treat 2026 as the year to prepare, because 2027 will reward the providers who moved early.

Just a final thought

Australia also needs a faster national response to AI-driven change in the workforce. AI is already changing tasks and workflows in real workplaces, yet the national skills framework development is moving too slow. The national training framework still has limited coverage of the skills and knowledge people need to work safely and effectively with AI. The cycle for updating training products is not keeping pace with how quickly change is happening. That leaves industry and RTOs filling the gaps while demand is shifting now. We need a stronger national effort and faster development and updating of training products to support targeted upskilling. When government decides something is urgent, it can move quickly. We have seen this in the past. Rather than treating AI literacy and skill development as a long consultation exercise, government should commission a focused, time-bound review of existing units of competency to identify where AI-enabled work practices need to be built into existing units, and where new units are required. If we do not catch up, we will miss the opportunities of this once-in-a-generation change. This requires clear direction and urgency from the Minister and agencies responsible for workforce planning and training product development.

Good training,

Joe Newbery

Published: 25th February 2026

Disclosure: For transparency, Joe Newbery holds an ownership interest in Unith AI mentioned in this article. Reference to Unith AI is included as an example only.

Back to Articles