Why ANZ's moving to consciously centralised software development
Nicky Darwin explains the benefits of moving from federated software development to a conscious centralised model
My experience on the tools still informs many of the decisions I make today. I used to roll my eyes when the business would start a sentence with ‘why can’t you just…’ or ‘how hard can that be?’
I’m not saying it’s always possible to entertain all suggestions, or that you can just hack and rack it, but they were right. Delivering to the outcome in the shortest time is actually what matters most. Especially with the pace of change at the moment.
With that mind shift I now look for what’s available, what can be reused, and focus on what matters. Right now I’m focusing all my energy on helping our engineers embrace further automation and leverage new technologies like AI so they can continue to produce great work in the most efficient way.
A great example of this is the Prism module within the new AI-enabled testing app we built called TARA. (It’s launching in July and there’s more on this below...)
Software development at ANZ
ANZ operates under a federated development model to accommodate the different teams working on different products for different customers.
The need for teams and individuals to choose their tools and processes, within a defined set of parameters and with appropriate governance, is where we need to be so we can get on with building the apps to help secure ANZ’s and our customer’s digital futures.
When I think of the impact this will have on our output, industrial England springs to mind (stick with me). At the time there were lots of factories building innovative machines, but they all made their own screws. This meant repairing machines required custom parts, parts weren’t interchangeable between manufacturers, and inventory management was complex and expensive.
When Joseph Whitworth introduced a standard screw thread in 1841, the ‘British Standard Whitworth’, it changed everything. Manufacturers were able to speak the same language and interoperability and efficiency in manufacturing improved greatly.
Back to the present day... Now that the foundational work on our main apps is largely done, we need to shift to an accelerated release cadence that better utilises available technologies and helps us to deliver more value more often.
Our ‘industrialised’ future
Engineering at ANZ is moving to what I call a ‘conscious centralised’ model. I call it ‘conscious’ as we’re not looking to mandate everything, only those things that don’t directly help us to solve the problems we’ve been challenged to fix.
The goal of this approach is getting to a point where engineers can meet our controls, assurances and standards by default, and not have to invest any of their time and energy working on core admin tools.
Conscious centralisation for the organisation means we can capitalise on the benefits of AI at scale and focus more deeply on the products we deliver. Our overall risk profile will be reduced, the security of our products will be enhanced, and we’ll be able to deliver value to the organisation more quickly and more often.
What I also like about this approach is there’ll be an even stronger alignment between teams, work will get done quicker, and engineering can move forward as a collective with no one being left behind.
Our approach to centralisation
Our approach to centralisation revolves around identifying specific areas that are core to software delivery and delivering targeted value quickly.
We’re deliberately not going full-in and enterprise-wide for everything, but rather consciously focusing on automated management of code, build, test and deploy that are available to everyone and that can deliver large gains.
This will result in all engineers having easy access to and using pre-packaged bundles of compliant and vulnerability-free infrastructure from day one.
Our progress so far
The first step in our move towards a conscious centralised model – and ultimately the delivery of our Golden Path – is the introduction of the GitHub code repository.
Implementing this digital backbone was a core stepping stone, and now we’re all in there with our code we can start making the move to the same build and code patterns across the bank. This really does mean you build once and well for all.
Testing was the obvious next place to focus on for me as most people’s work passes through there and we have almost 1,900 testing engineers in the bank.
Accordingly, a team of engineers from the Quality Engineering domain have built an app that’ll improve the accuracy, efficiency, and scalability of testing across the enterprise. Equally, it will also lessen the need for license-based software and manual testing.
It’s called TARA, short for Testing Assistant in Research and Automation, and it houses advanced AI-powered testing modules that’ll revolutionise the quality assurance process. TARA is built on a Vertex AI Model Garden and utilises the Gemini Frontier Model.
So far we have nine tenants who are utilising all or some of the three use cases that are live:
Playwright Automated Test Generation automates the generation of test scripts for Web UI – including saving to GitHub, execution, and self-healing capabilities.
AskTARA is a conversational AI assistant that answers queries based on ANZ-specific context.
Super Test automates the generation of test scripts for APIs – including saving to GitHub, execution, and self-healing capabilities.
One of TARA’s strengths is that it can comprehend and respond to ANZ-specific text queries in a natural and contextually appropriate way using our Confluence and JIRA documentation. And in the future – should a player launch an app with similar capabilities – our functionality can quickly be transferred over.
Enhanced test optimisation with Prism
In July we’ll be rolling out TARA’s domain-agnostic Generative AI-powered test optimisation module, Prism. I think of Prism as a GPS for testers as it provides better clarity on the task at hand, possible next steps to take and oversight of the journey ahead.
When prompted, Prism will mine thousands of test cases across multiple platforms and then extract the key elements – actions, data, and validations – and transform them into structured, reusable assets that’re ready to go.
It’ll then takes things even further by responding to the query with:
Recommendations on optimised execution chains based on actions and flows that others in similar situations have taken, and
Visualise relationships through an interactive Test Case Canvas.
Considering how many thousands of manual test cases we’ve got going on at any one time, this is going to prevent a lot of duplication, simplify and streamline the testing process, and improve our speed to market massively!
What’s next?
We’ll continue to take steps towards transforming our software development into a structured, service-driven ecosystem, so that our engineers won’t need to manage infrastructure or build, code, test, and deploy software themselves.
Ultimately, we’ll see our engineers consuming compliant services that allow them to commit code on their first day, move seamlessly through the organisation, (as core tooling is standard), and shift from cutting-edge to core banking technologies with ease – all while maintaining the highest of standards!
It’s an exciting time to be building software right now, and even more exciting to be doing it at ANZ. I can’t wait to continue this journey with you and to celebrate as we deliver even more together in the future.
Nicky is ANZ’s Head of Engineering and Platforms. She is an experienced transformation leader within the banking and finance space who specialises in front office and data technology.
Nicky is experienced in leading from concept to delivery large transformation projects, regulatory reform and system integration.
She is skilled in managing multimillion dollar budgets and large scale cross regional resourcing.
This article contains general information only – it does not take into account your personal needs, financial circumstances and objectives, it does not constitute any offer or inducement to acquire products and services or is not an endorsement of any products and services. Any opinions or views expressed in the article may not necessarily be the opinions or views of the ANZ Group, and to the maximum extent permitted by law, the ANZ Group makes no representation and gives no warranty as to the accuracy, currency or completeness of any information contained.