Early adopters of MYOB Acumatica’s spring release have reported savings of 20 hours a month release thanks to AI-driven automation. Valantis Vais hopes that’s just the beginning as they embark on a journey to the ‘autonomous business’.

Vais, MYOB general manager for product, product marketing and design, told iStart mid-sized businesses across Australia and New Zealand are on the pathway towards autonomous business – albeit, only in the early stages of the journey – and the latest Acumatica release builds out capabilities that live “somewhere between level 2 and level 3” on the five stage journey.

“What really excites me is there are going to be almost business specific use cases that mid-sized businesses will identify to use this technology.”

Breaking it down, Vais at level 0, companies likely haven’t even digitised their processes and are still living in the paper world.

“From one to two they need to start using automations and no code/low code and then AI and so on and so forth, all the way up to multi-agent AI’s which are more in that level four/level five.”

The ultimate endgame? “A business where the people are able to focus on the things that derive value for their customers and less on some of the back-end activity that needs to occur for them to deliver that value.”

Getting practical

The new release, he says is about surfacing practical AI capabilities that help manage and reduce some manual tasks.

That word – practical – is doing a lot of heavy lifting here. Vais notes that while there’s a very high level of curiosity from midsize businesses about AI, they’re wary of hype and sunk costs, and want tools that provide practical benefits in the here and now rather than promises of a sci-fi future.

A recent MYOB survey of 500 local mid-sized business leaders found that across the board payroll and workforce management is the top area cited for automation in the next 12 months, highlighting a readiness to reduce ‘busy work’.

“The pattern in our findings is consistent: Mid-sized businesses want to target routine, high volume work first to free skilled teams to focus on throughput, quality and customer commitments.”

The headline acts in the release, currently rolling out across Australia and New Zealand, are new accounts payable bill entry powered by OCR and semantic interpretation, expense management with AI-driven receipt capture, AI Advisor anomaly detection and Auto Complete.

The first two are potential game-changers for finance teams. AP automation pulls invoices from email inboxes, using OCR to read the documents and semantic interpretation to ‘read’ them and infer and extract information such as supplier names and stock items, enter the data into the ERP, and produce an AP bill based on the information, ready to be reviewed by a human.

It’s here the 20 businesses on the MYOB Early Access Program have saved up to 20 hours a month. MYOB modelling predicts up to 60 percent time savings for AP invoicing/reconciliation with AP automation enabled, plus additional savings across reporting and reconciliations.

Expense management works similarly: Employees snap a photo of a receipt, and the system pre-fills entries, categorises items and matches transactions to projects or customers.

While there’s no benchmarking or test results on the accuracy of the OCR, Vais says MYOB has seen acceptance rates upwards of 90 percent. Over time the system continues to learn with a built-in workflow to train the system.

As MYOB’s plans expand, the company is also investigating multimodal ‘receipt recognition chat’ and assistant experiences.

Vais is also excited – though for different reasons – about an AI advisor for anomaly detection.

“On the automation side, the value is extremely clear. Essentially, we’ll remove a lot of mundane work for financial controllers or people who need to submit expenses,” he says. “But on the anomaly detection side, I’m excited to see the use cases mid-sized businesses will come up with.”

The anomaly detection capability sits across the entire platform and enables users to look for anomalies in any data tables.

“Because it has such broad application across the platform, there are specific use cases that we can see people using it for – on your general ledger to identify when a specific transaction has been allocated to the wrong account, for example, so it helps you close your books more accurately and faster.”

It also opens the doors for more sophisticated means of pursuing someone who hasn’t paid in a timely fashion, ensuring you target those who have the most anomalies.

“But what really excites me is there are going to be specific, almost business specific, use cases that mid-sized businesses and partners will identify to use this technology.” Those use cases are likely to include project profitability and inventory stock outs, among other things.

A raft of other features have also been added across the key capability areas of finance, people, sales and customer (including streamlining of address management), field services (with improved, automatic, tracking of specific serial or lot numbers) projects, supply chain and production.

Foundations first

But before anyone gets carried away with visions of AI utopia, Vais offers a reality check: Clean, structured data is non-negotiable.

“All of the models are reliant on high-quality data. Garbage in, garbage out still applies,” he warns. Businesses need to ensure their data is accurate, accessible and in the cloud. “It’s almost the ticket to play.” Without that foundation the autonomous business remains a pipe dream.

And while AI can automate tasks, it can’t automate trust. MYOB’s early testers were cautious about the accuracy of the OCR, and whether it would surface items with issues. That saw them double-checking system generated entries.

Trust in the system built up ‘relatively quickly’ Vais says.

The workflow design helps here. “As AI starts to automate more and more, it becomes critical for that workflow that’s using AI to help the human have confidence that it’s going to surface up the areas that require intervention, otherwise the user will intervene on everything and have no time savings, or intervene on nothing and potentially have errors.

“It’s a Goldilocks situation, where you need to give them the right visibility so they can intervene where the model isn’t accurate or is unsure.”

Future forward

The spring release is just an opening act. Vais says MYOB will deep automation and introduce more intelligent capabilities across Acumatica as it moves customers towards that autonomous business vision.

In the New Year, it will launch AI Studio, giving businesses the ability to create their own AI agents to further automate workflows.

Conversational agents to query your ERP in plain language – ‘what’s my best selling product this month?’ ‘what are the products this customer has purchased in the past?’ – rather than digging through dashboards, are also on the horizon.

The advisory layer will expand too, with cross-sell recommendation, designed to surface revenue opportunities during customer interactions.