Dr Amantha Imber says organisations must shift from adoption to workflow redesign to unlock value
AI adoption is no longer the hard part for many Australian workplaces. The harder question is whether all that activity is actually changing how work gets done.
ELMO Software's 2026 HR Industry Benchmark Report illustrates the gap. Some 93% of Australian HR professionals use AI for HR tasks at least occasionally, yet only 19% say AI-enabled tools are extensively integrated into day-to-day workflows. Just 15% describe its impact on their HR function over the past year as transformative.
At a recent ELMO customer event, organisational psychologist and AI consultant Dr Amantha Imber argued the next stage of workforce transformation is redesigning work around AI and getting more rigorous about value.
AI is moving faster than workflows
For the past few years, employees have been encouraged to experiment, learn how to prompt and find ways to save time. ELMO's latest Employee Sentiment Index shows speed is the biggest reason Australians choose AI (57%).
But completing an existing task faster isn't the same as redesigning the work itself, and it doesn’t always lead to positive employee outcomes. Almost three-quarters (72%) say when AI lets people work faster, more is expected of them, while 49% have felt uncomfortable with its use at work.
For Imber, there are three questions every executive team should answer: “‘What is your why for AI?’, ‘what will you do with the time saved?’ and ‘what are your success metrics?’” she says.
“When thinking about your people, question two is critical. Silence breeds fear of job cuts.”
HR needs a seat at the redesign table
Imber argues organisations getting AI transformation right won't leave it solely with the technology team.
“The CEO needs to fully endorse the transformation and model AI leadership, with genuine co-ownership between technology and people leaders.”
ELMO's research suggests that model is far from universal. Some 39% of Australian HR professionals say IT or Technology is ultimately responsible for AI adoption and transformation, while 19% describe responsibility as shared between HR and IT.
Yet many of the biggest questions AI creates are workforce questions: which tasks should change, what capabilities people will need, how saved time should be redirected and how roles may evolve.
When more output isn't better output
Then there is quality. “AI slop”, shorthand for generic AI-generated material online, is facing growing and necessary scrutiny. Take LinkedIn, which has introduced a “Seems like AI slop” feedback option.
The same problem exists inside organisations. Think of the employee receiving pages of indecipherable AI-generated feedback from their manager that takes longer to make sense of than the original task itself.
ELMO's HRIB research supports this, with 32% of workers reporting excessive rework after using AI, 31% having difficulty validating outputs and 29% experiencing inaccurate or unreliable results.
“AI makes volume effortless, so volume is no longer a signal of quality,” Imber says.
“If your organisation is still only measuring performance with metrics like licence adoption, daily logins and token usage, it's time to regroup. Time saved is only useful when it is paired with a quality measure.”
US software business Clay recently introduced a company-wide AI writing policy built around principles including “writing is thinking” and rejecting the idea that longer means better.
Imber applies a similar principle to her own work.
“I now publish ‘time to write’ alongside my Substack posts. It forces me to monotask and demonstrates to the reader the effort involved,” she says.
“I encourage leaders to think about measures like this for their own teams as a quality proxy.”
Redesign one workflow before chasing the next tool
Financial discipline will also increase as organisations move from experimentation to agents and automated workflows. That brings tokenomics into the conversation: understanding and managing the cost of AI consumption as usage scales. As this happens, finance leaders will increasingly expect that spend to translate into better outcomes.
“Most people cannot unpack a workflow, which is the prerequisite skill for agent design,” Imber says.
“A practical starting point is to choose one repeatable workflow, define what a good outcome looks like, break the task into clear steps and decision points, then test the agent and measure both time saved and quality.”
Even asking team members to share “how I used AI this week” can surface ideas and help people learn from each other. That matters when 43% of employees say they aren't clear about when AI should and shouldn't be used at work.
The aim is not to automate everything quickly, but to build the capability to identify where AI genuinely improves work.
“Reliable agents take thoughtful building and testing,” Imber says. “A ten-minute build is a red flag.”
For HR leaders, the measure of progress is no longer how much AI an organisation uses, but whether it is producing better work.
ELMO's five-minute AI Maturity Assessment benchmarks HR teams on AI readiness and effectiveness and provides a tailored action plan for building AI capability. Take the assessment here.
This article was produced in partnership with ELMO
