AI for manufacturers: Foundations, real-world examples and what’s next

Sponsored

There is a lot of noise and many wild assertions attached to the AI-for-business conversation, but definitions matter.

Valantis Vais, General Manager – Product and Product Marketing at MYOB, prefers to separate AI into three practical categories.

First is machine learning, or traditional AI, for example a pick-and-place robot on a line able to identify an object, or object recognition for documents, based on educating a system based on many examples.

Second is the subject of the current wave, large language models (LLMs), which underpin products such as OpenAI’s ChatGPT, Anthropic’s Claude and others, which open up new ways to search, summarise and interact with data. 

The next wave is agentic AI, says Vais, which is based on LLMs, though “what they're able to do is actually do more end-to-end automation. So they're able to actually complete a task from start to finish.”

Vais believes agentic AIs will be useful in scenarios such as training a robot a lot easier than in the past.

“If we then think about agentic experiences in the office, think of these as being things like a procurement agent, and it tracks component usage, predicts shortages, can propose or even auto-create purchase orders with pre-approved guardrails,” he adds.

“And then it may actually route back to the human to approve it. So that's kind of an agentic experience that you can imagine operating in an office.”

Different surveys show different results on Australian workplace adoption of various kinds of AI.

Results from a Reserve Bank of Australia survey of 100 medium and large enterprises were published this month. This found that nearly 40 per cent displayed “minimal” use of AI, just over 20 per cent were “moderate” in their adoption, and 10 per cent had embedded it into more advanced processes.

A recent MYOB survey of midsize businesses found that, over the next 12 months, 45 per cent of respondents plan to adopt solutions for payroll and workforce management, with 43 per cent looking at inventory and supply chain optimisation, and another 43 per cent targeting automation of administrative tasks and repetitive backend processes.

Besides the three waves of AI washing through businesses, another concept is mapping a business’s level of sophistication in terms of Level 1 to Level 5, long used to categorise levels of autonomy in cars.

At the bottom is Level 0. Think of a car operated entirely by the driver, or a company with no digitisation and paper-based systems. At the top? A completely autonomous vehicle.

A company at level 1?

Vais says this looks like, “A scenario where you've digitised your activities. Level two, a degree of automation. Level three, the injection of these agentic AI experiences that I talked about earlier.

“Level four, multiple agents interacting with each other, all the way to level five, where you actually have a business that runs itself with respect to the back office, but then essentially allows the people to do the activity that adds value.”    

For Australian manufacturers, the safest AI gains will come from targeted use cases and disciplined change. MYOB Acumatica centralises operational and financial data, applies role‑based controls, and maintains established approval paths so improvements land with assurance, not disruption. Learn more here.

Episode guide

0:53 – An introduction to our guest.

1:50 – What do we mean when we talk about AI in business.

2:58 – Traditional AI or machine learning e.g. vision systems.

3:32 – Generative AI or LLMs. The current wave.

4:14 – Agentic AI. The future. Also leverages LLMs, but able to complete end-to-end automation.

5:10 – Examples of agentic AI experiences in the office.

6:05 – Where is AI making a difference to productivity.

6:52 – Mapping autonomy in business onto the different levels of autonomy for cars and how it could play out over time.

8:16 – Where AI is providing benefits currently, such as in consistent and accurate documentation and improved administration.

9:20 – How companies can make use of the huge amounts of data that they currently collect or could be collecting.

10:15 – Profitability challenges and how AI might come up with answers.

10:50 – Leading applications from those surveyed include in payroll, inventory and supply chain optimisation and reducing administration tasks.

12:05 – The starting point is a foundation of digital data, stored in the cloud, and going business-first rather than technology-first.

13:20 – You should also consider the emotional aspect and how people understand their use of AI.

14:40 – Instead of waiting for shopfloor reports, line-level reports can be summarised and updated regularly.

16:05 – Top challenges for adopting cloud. The top answers were lack of inhouse expertise, high and unpredictable migration costs, and ongoing operational costs. The answer might be starting with pilots.

17:40 – Work with partners who understand your business’s “from” and “to”.

19:35 – What 2026 could deliver for businesses' AI adoption.

21:46 – MYOB’s Acumatica’s three pillars.

22:38 – A message to those on the fence: there are no shortcuts and here’s why.