Published on the 21/08/2025 | Written by Heather Wright

ERP info, Pronto!…

Chad Gates, managing director of Australian ERP provider Pronto, has a straightforward goal with AI and agents: He wants to help lift customers ‘out of the busy noise’ and enable them to turn their workforces into ones that look much bigger than they actually are.

The company, which is a long-time IBM partner, has signed an agreement with the global giant enabling integration of agentic AI capabilities into its Pronto Xi ERP platform, via Watsonx.

“We have agentic capability around inventory, purchase ordering and sales orders that we will be looking to release in the next few months.”

The deal will see Pronto’s AI infrastructure complemented by IBM’s agentic AI tools, with Gates telling iStart that Pronto customers will be seeing plenty of new AI and agentic capabilities rolling out over the rest of this year.

Among the key things customers say they’re looking for is being able to do more problem solving and troubleshooting within Pronto, finding answers more easily.

“That’s a key one which lead us down the path of our IBM arrangement where we built a Watsonx-based support chatbot, essentially, which has ingested over 50,000 pieces of documentation and information about versions of Pronto going back 15 years, which we will shortly be deploying to all Pronto customers,” Gates says.

Other use cases include long-range financial forecasting, predictive analytics around inventory – inventory maximisation is a big goal – and purchasing, using inputs such as stock history and external factors.

Finding anomalies in inventory categories, particularly for those in a franchise arrangement or similar, is also on the cards, while a tool to support engineer allocation in customers has been built.

So where does the current hot tech offering of agentic come in to play?

Pronto is using IBM Orchestrate, which is an agentic capability in the Watsonx stack and other capability it has built inside Pronto to develop agents, particularly around surfacing information more easily for end users.

“For them to do processes might have required diving through screens, following process diagrams, all those sorts of things. Being able to front those processes with an interface and LLM – whatever that LLM happens to be – and being able to have natural language conversations and say ‘Give me the history of all these inventory items over the last six months, but take into account seasonality and supplier lead times and show me an analysis of what I should be reordering’, and then deploying an agent to raise those orders or update data in Pronto. That’s pretty interesting.”

The company is also working on a capability that could potentially automate the ability for bugs to be self-diagnosed and self-repaired automatically as a customer logs a call.

The agent would check if the bug already has a fix, and if not raise a ticket in Pronto’s back end, go and diagnose the bug ‘potentially’, and suggest a fix.

“There are a lot of things that can potentially be automated to help elevate the humans above the busy tasks that have taken them significant amounts of time in the past.

“We are coming up with agentic use cases on an almost daily basis and already have a fair bit of capability around inventory, purchase ordering and sales orders that we have worked on in the back-end and will be looking to release in the next few months,” Gates says.

“By the end of the calendar year, we will have a whole bunch more agentic capability that we will be releasing into the Pronto Xi 780 version, which will include the LLM frontend chatbot inside Pronto itself, as well as a mobile application that provides the same sort of access to Pronto.”

Gates is keen to see intelligence democratised across businesses, enabling companies to develop the capabilities of their teams.

He notes that while AI often gets a bad rap as a technology that is stealing jobs, for small businesses that’s not such a concern.

“They have the opposite problem and want to be able to get more out of the people they do have and be able to more strategically invest,” he says.

Gates says Pronto has been working with customers and IBM to develop proof of concepts for several years

“We wanted to take a really practical approach to AI,” he says, noting that customers, while keen to understand AI and agentic are often not quite sure what it means for their business, where to start, how it can benefit them and how to go about using it.

“Our approach has been very much about creating an environment that creates practical business outcomes but with a very solid framework of trust and governance around it.”

“Like every business, we have been doing a lot of work in the background, sorting out our own capability, working out where we invest, investing significant amounts into Pronto cloud in terms of hardware, and deciding what we want to own, who we partner with, what technologies we use and what our use cases are and what will actually bring value to customers based on what they are telling us.

“It is very much about business outcomes, rather than just taking the lazy approach and putting widgets in place and say we have AI.”

With ERPs home to critical company data, security and governance have been critical, Gates says.

“Where the value lies for AI is in the data it accesses, and AI is really good at surfacing information. The question is how do you make sure it doesn’t surface information the end user isn’t meant to see?

“You don’t want staff asking chatbot what manager gets paid. That would be bad!”

Gates says Pronto has built security layers into the API layer and is using MCP (Model Context Protocol), an open standard that defines how AI applications can connect to external tools and data sources.

“In those you can tell the AI what a user can and can’t see. So we build multiple layers of security around the user experience so the AI knows its boundaries, the back end knows the boundaries it’s allowed to give the AI and we have full transparency of all the interactions the AI is having in the backend.”

Governance tools enable data such as PII or telemetry to be stripped out, for example in the case of the mining support calls customers make and using that data to assist other customers to resolve their support issues, ensuring that data isn’t accidently shared.

The LLMs are also hosted inside Pronto Cloud with data kept within Pronto’s firewalls and within the sandbox created for customers.

“This means there is no danger of that data escaping into a publicly available large language model for example. So they can safely query that information and it will be kept inside our ecosystem.”

Gates is bullish about agentic AI’s impact on ERP, saying it will fundamentally change the way end users interact with an ERP system.

“We all know ERP systems are transaction engines so they generate a lot of data, they have a lot of complex processes – one of the reasons use ERP is because it does hard stuff so you can do hard things well, and it has good data integrity, history and authenticity to it.

“AI enables users to interact with the system in a way that is very natural to them.”

Gates says he sees a future where for some functionality ‘we may not even build screens at all’.

“It might be that the only interface to that functionality will be an AI interface and it might be whatever AI interface you happen to be using and it potentially draws interfaces on the fly as an end user request them – so show me information in this format, write a report that looks like this.

“The ability to surface potentially what is complex processes and make them very accessible and easy to interact with is certainly part of future for ERP without a doubt.”