ARM Hub at CEDA’s AI Leadership Summit

ARM Hub joined leaders from business, government and academia at CEDA’s AI Leadership Summit at ICC Sydney last week, with a stand in Research Alley on day one and a workshop on day two.

On Tuesday, 27 October, the ARM Hub team was in Research Alley sharing the latest on Digital Transformation in the Age of AI, along with summary reports from the University of Technology Sydney’s Turning AI into Productivity report. Special guest Roy the Robot joined the stand with the team from UTS.

On Wednesday, 28 October, ARM Hub’s Charlotte Birkinshaw and Communications Director Mike Woodcock joined the National AI Centre’s Emma Haller and SMEC AI’s Andrew Lai and Krish Parekh for Beyond the MVP: How to scale AI securely and deliver real business value. Hosted by IBM and facilitated by SMEC AI and ARM Hub, the workshop focused on practical implementation, production deployment, cybersecurity, scaling and ROI.

The workshop was built around a common problem: AI pilots are easily built, but scaling successfully is always the challenge.

Charlotte walked the room through five tests for deciding whether a pilot is ready to roll out, and they’re useful for anyone considering the next stage of AI adoption:

• Painkiller or vitamin? Only scale AI that solves a real problem, such as cutting search time, rework or reliance on one key person. If you turned it off tomorrow, who would complain? If nobody would miss it, don’t scale it.

• Design for the people using it. Before building anything, define the outcome that matters, what good output looks like, how people will trust it and who owns the process. Adoption has to be designed in from the start.

• Match the tool to the task. AI earns its place with messy language, judgement and variety. Predictable rules suit automation, which is cheaper, repeatable and easier to test, and a broken process needs fixing first. Often the answer is both: AI for the words, automation for the rules.

• Judge it on production numbers. An example pilot showed 70% faster drafting in testing, and 12 of 15 pilot reps were still using it. But one in 10 drafts had an error, and the whole thing ran on manual CRM exports and personal AI accounts. Real adoption, review time and running costs show whether it’s ready.

• Secure it from day one, then start small. Production needs an enterprise AI service with agreed data terms, role-based access, prices pulled from the price book instead of generated, and human approval before anything reaches a customer. From there, start with a narrow scope, let AI draft while people decide, and automate later.

The five tests draw on what ARM Hub’s AI Adopt Centre hears in its workshops with businesses, and they apply just as well to organisations still looking for the right use case. Businesses ready to take the next step with AI can get in touch with the team at armhub.com.au/join-us.

Thanks to Sheila Sutjipto at UTS for helping organise Roy’s visit.