Software development is undergoing a massive shake-up with AI agents assuming aspects of the development process for faster and more cost-effective modernisation of legacy systems and the creation of new digital platforms.
Joel Macfarlane, Datacom’s Director of Data & Engineering, recently outlined how AI agents are now part of Datacom’s software development teams, working alongside human engineers and even collaborating on kanban boards.
Over the last few months, Datacom customers across Australia and New Zealand have experienced our AI-powered application modernisation demos and the response has been overwhelmingly positive, says Sunny Katira, Managing Director - Professional Services, Australia and New Zealand at Datacom.
“We’ve got up to 70% of code being written by agents on various projects. It’s entirely possible,” says Katira.
“We’re doing it for one of our largest public sector customers in Australia. This is not research and development or experimentation. We are implementing this in real-life production scenarios.”
This shift means that what once took months of manual coding can now be achieved in a fraction of the time. Customers who may have previously hesitated to invest in application modernisation due to cost or uncertainty about return on investment (ROI) are now seeing the business case stack up.
“Telling customers that AI can actually cut your development time in half is a great statement to make,” Katira says.
“But then they want to know how and what. What our demo does is show exactly how we craft AI agents. We have a team of agents that interact with each other and write code. That’s the ‘how’.”
Across New Zealand and Australia, organisations are sweating their IT assets, reluctant to commit to modernisation programmes in tough economic times.
While the immediate benefit for many customers is cost reduction, up to 50% in some cases, Katira is adamant that the true value of AI agents goes beyond just savings.
“The legacy-to-modernisation conversation is definitely a cost-saving thing. But crafting new software is how you gain a competitive advantage,” he says.
AI agents don’t just make it cheaper to refresh old systems. They make it possible to build entirely new digital products and services that would have been out of reach due to budget or skills constraints.
Datacom is using AI agents for business analysis, developing requirements, testing, and the software coding itself, with human staff increasingly overseeing agents and checking their work. It allows for rapid development of new software and platforms, more cycles of testing and higher quality, for purpose products.
A less understood but increasingly critical dimension is the role of hardware, particularly graphics processing units (GPUs), in AI-powered software development.
In the past, software performance was largely hardware-agnostic. Code would run similarly whether deployed in the cloud or on-premises. Now, with AI workloads, the quality and capability of the underlying GPU make a tangible difference. High-performance GPUs enable software to perform complex calculations faster, with lower latency and at a lower cost. This is especially crucial for AI-powered applications that require supercomputer-like processing.
“With GPU-based AI processing, the underlying GPU actually makes a tangible difference to how well the software code performs,” says Katira, who points out that many AI workloads on public cloud platforms require optimised GPUs and are processed offshore.
“So having the right partnerships and infrastructure makes a big difference”.
Datacom is investing in local GPU infrastructure in its data centres to ensure that AI workloads can be processed within New Zealand and Australia, addressing both performance and data sovereignty concerns.
“It’s in country, low latency, high performance, fully sovereign,” Katira points out. He predicts that software will increasingly be tailored to take advantage of specific AI hardware, with agents optimised for certain GPU platforms, mirroring the rise of platform-as-a-service models in cloud computing.
The next frontier, according to Katira, is the emergence of AI as a platform. Companies like Nvidia are releasing agent blueprints optimised for their GPUs, much like how Salesforce or Microsoft offer platforms for business applications. In this model, the software or agents may be freely available, but the real value and differentiation come from the underlying hardware and how it is leveraged.
Datacom’s innovative approach to deploying AI agents in the software development process offers immediate savings for those pondering application modernisation projects. But the agents also unlock new value by making it faster and more efficient to develop new apps from scratch and optimise them for the age of AI.
Our expert engineers can help your organisation modernise its legacy systems. We're using the latest technologies and best practices to deliver high-performing, scalable, secure and user-friendly solutions - while reducing costs and project timelines.
We can help your organisation to modernise your applications, increase resilience and become equipped to succeed in the complex, rapidly changing business environment that we are all operating in. Our approach can help your team find its competitive advantage and deliver value to your customers and stakeholders.
