AI Adoption Success in Australia
AI at Work in Australia: Success and Productivity versus New Business Creation
Prepared by Rob Leach July 2025
Executive Summary
Market Position
Australia has moved from AI experimentation to measurable deployment. Fifty-two per cent of businesses now use AI, with services firms leading at 56 per cent adoption. Yet most treat AI as an efficiency tool rather than a revenue generator. Only 23 per cent prioritise technology adoption whilst 42 per cent focus on new business development.
The productivity gains are substantial. Commonwealth Bank achieved 30 per cent fraud reduction and 40 per cent faster call-centre resolution. Telstra cut service creation times from 12-18 months to three months whilst reducing processing time by 30 per cent. Ashurst's controlled trial showed 45 per cent faster briefing drafts with 2.5 hours saved per document across 400 lawyers.
The Innovation Gap
Australian enterprises capture efficiency gains but lag in revenue creation. Global leaders demonstrate the potential: L'Oréal reduced product development cycles by 60 per cent and increased conversion rates by 22 per cent. Insilico Medicine moved a drug candidate from discovery to phase-2 trials in 30 months—half the normal timeline. These examples show AI's capacity to generate new business models, not just operational improvements.
Strategic Requirements
Success demands enterprise-wide commitment. Norway's sovereign wealth fund mandated AI across all units, achieving 20 per cent productivity gains equivalent to 213,000 saved hours. CEO Nicolai Tangen's directive that non-adopters would not be promoted removed implementation ambiguity. The fund built secure infrastructure and maintained human oversight—traders cannot execute AI-generated trades without approval.
The organisational challenge exceeds the technical one. Seventy per cent of implementation difficulties stem from people and process issues, with only 30 per cent attributed to technology. This explains why fewer than 20 per cent of organisations possess the data strategies and governance frameworks to scale AI effectively.
Risk Profile
Three critical risks require immediate attention. Shadow AI usage threatens confidentiality—31 per cent of lawyers use unauthorised public tools despite formal deployment reaching only 18 per cent of firms. Hallucination creates negligence liability, with 70-80 per cent of professionals fearing erroneous outputs. Skills gaps compound these risks, as 64 per cent of professionals lack formal AI training.
Competitive Dynamics
Early movers secure sustainable advantage through superior margins and client value. AI-enabled rivals can undercut traditional pricing whilst preserving profitability, creating pricing pressure that late adopters cannot match. Fifty-eight per cent of in-house counsel expect AI efficiencies to reduce external provider reliance for routine work. This shifts competitive dynamics permanently.
Implementation Framework
Success requires board-level strategy spanning governance, technology deployment and workforce transformation. Enterprises should centralise procurement to eliminate shadow usage, invest heavily in training and change management, and focus on high-impact use cases with measurable benefits. Human oversight remains essential—AI augments rather than replaces expert judgement.
The most effective organisations create internal "enabler" teams and ambassador networks to support adoption. They build secure AI platforms, establish baseline metrics and track improvements rigorously. Without measurement, initiatives risk becoming experiments without impact.
Financial Returns
Early adopters demonstrate compelling returns. The case studies show ROI ranging from 21 per cent to 374 per cent, depending on implementation scope, with payback periods between 10-28 months. These returns reflect both direct productivity gains and avoided opportunity costs from competitive displacement.
Regulatory Opportunity
Australia's fragmented regulatory landscape creates positioning opportunities for proactive organisations. Courts issue separate practice notes whilst lacking coherent frameworks. Early movers can influence emerging standards through white papers and industry submissions, establishing thought leadership whilst competitors await guidance.
Strategic Imperative
Australian enterprises face a narrow window to capture AI advantage before competitive dynamics crystallise. The choice is between leadership through early adoption or marginalisation by AI-enabled rivals with superior cost structures. Given productivity gains of 30-90 per cent and quality improvements reaching 40 per cent, the economic case for immediate action is unambiguous.
The most successful organisations will combine operational efficiency with new revenue creation, using AI to both defend existing margins and discover untapped opportunities. This dual approach—productivity enhancement plus innovation—separates market leaders from efficiency-focused laggards.
Decisive action now positions organisations as AI-enabled market leaders rather than traditional operators under permanent pricing pressure.
Full Report
1. The difference between AI for productivity gains and AI for new business creation
Artificial intelligence can be deployed for two distinct outcomes. Productivity-enhancing AI improves the efficiency of existing processes or services, while innovation-driven AI creates new products and business models. Surveys by the Australian Industry Group (Ai Group) show that about half (52 per cent) of all businesses are already adopting AI, with services firms (56 per cent) leading industrial firms (38 per cent) in uptake. Despite this broad uptake, most companies still treat AI as a tool to improve operational efficiency rather than a platform for generating new revenue. Fewer than one in four (23 per cent) Australian firms list technology adoption as a strategic priority, whereas 42 per cent prioritise new business development.
Globally, the story is similar. While enterprise adoption of AI is widespread, with some studies showing over 80 per cent of companies using or experimenting with the technology, a significant 'readiness gap' persists. Multiple analyses indicate that a small fraction of organisations—in some cases fewer than 20 per cent—possess the mature data strategies, IT architecture and governance frameworks required to scale AI effectively and generate enterprise-level value. These gaps underline why many projects never progress beyond the pilot stage.
AI for productivity focuses on automating repetitive tasks, summarising information and assisting human experts. Examples include summarising customer notes for call-centre agents or automating document processing in claims management. The value comes from freeing people to perform higher-value work and reducing costs. In contrast, AI for new business creation uses machine-learning models to design new products, enable personalised services or open entirely new markets. Generative models can brainstorm novel formulations, such as L'Oréal's AI systems that reduced product-content development cycles by 60 per cent and increased conversion rates by 22 per cent. In biotechnology, Insilico Medicine's platform combined target discovery and molecule generation to discover a drug candidate that moved from project inception to phase-2 clinical trials in under 30 months—about half the usual timeline. These examples show that AI is not only a back-office tool. It can generate new revenue streams when deployed creatively and responsibly.
2. Learning from success: Norges Bank
Norway's sovereign wealth fund offers one of the clearest demonstrations of enterprise-wide AI adoption. Norges Bank Investment Management (NBIM), which manages roughly US$1.8 trillion, mandated AI across all business units in 2024. CEO Nicolai Tangen appointed a six-person "AI enabler" team, set up AI ambassadors in each department and declared that employees who refuse to use AI would not be promoted.
The fund built its own secure AI platform and gives portfolio managers access to Anthropic's Claude model to query market data, summarise earnings calls and monitor news in 16 languages. Within months, NBIM reported about 20 per cent productivity gains—equivalent to around 213,000 hours of work saved. A generative AI trading model uses machine-learning algorithms to predict short-term equity returns and to net trades internally, reducing transition costs and improving returns. These gains were achieved without increasing headcount.
NBIM's experience underscores several lessons for Australian leaders. First, success begins with clear direction from the top: Tangen's public commitment removed any doubt that AI adoption was optional. Second, secure architecture matters. NBIM built a private AI platform to protect data and comply with regulations. Third, AI remains a tool for augmenting rather than replacing human judgment. NBIM's traders cannot execute AI-generated trades without human approval. Fourth, workforce capability and engagement are critical. The organisation invested heavily in training and created a community of ambassadors to help colleagues develop new skills.
These practices align with broader industry analysis, which suggests that the greatest challenges in AI adoption are organisational rather than technological. Some studies indicate that up to 70 per cent of implementation challenges stem from people- and process-related issues, with only 30 per cent attributed to the technology itself.
3. Australian success stories
Australia is moving beyond experimentation to deploy AI at scale across multiple sectors. The following case studies demonstrate how enterprises are realising productivity gains and creating new value.
Carsales — computer vision to drive listings
Carsales.com Ltd introduced an image-recognition system called Cyclops. When private sellers choose Cyclops on their mobile app, the model automatically selects the best angles for vehicle photographs. This AI feature delivers an 11 per cent higher advertisement completion rate compared with manual photo uploads, with an accuracy rate of 97.2 per cent. The improvement boosts both customer experience and operational efficiency.
Urbanise — AI for facilities management
Property management firm Urbanise.com Ltd uses machine-learning models to automate transactions and provide contextual analytics. Its AI-enabled platform contributed to 193 per cent growth in facilities-management revenue and 110 per cent growth in strata recurring revenue over three years. These gains show how AI can generate new income in traditional service industries.
Rio Tinto — autonomous operations and predictive maintenance
Mining giant Rio Tinto operates the world's largest autonomous rail network in Western Australia. Its AI-driven Mine Automation System now covers 98 per cent of mining sites and has halved Tier-1 safety incidents. In addition, a generative AI classifies faults in autonomous locomotives.
Commonwealth Bank of Australia — AI-enabled banking
CBA has integrated AI across customer protection, messaging, lending and developer productivity:
Fraud and scam reduction. In late 2024, the bank reported a 30 per cent drop in customer-reported fraud, thanks to AI measures. The system processes over 20 million payments per day and sends 20,000 proactive alerts per day, with plans to increase to 35,000.
Customer service. AI-powered messaging reduced call-centre wait times by 40 per cent. The bank's customer engagement engine makes around 55 million decisions each day, enabling staff to have more relevant conversations.
Loan processing. AI-powered document intelligence is being deployed to accelerate core banking processes. In pilots, the technology has achieved automation and accuracy rates of 50-85 per cent on varying document types for customer verification, and is expected to reduce the time for complex annual business credit reviews from approximately 14 hours to just two.
Employee productivity. CBA has rolled out Microsoft 365 Copilot to 10,000 active users. Of these, 84 per cent reported they would not want to go back to working without it, with early adopters saving an average of 16 per cent of their time on repetitive tasks. The bank also deployed GitHub Copilot for its technical teams, where engineers adopt approximately 30 per cent of the AI-driven code suggestions, accelerating development processes.
Benefits Finder. The AI-driven 'Benefits finder' tool in the CommBank app has connected personal and business customers to an estimated A$1.2 billion in government grants, rebates and concessions since its launch in 2019.
These initiatives illustrate how AI can simultaneously improve customer safety, accelerate lending and boost developer efficiency.
Telstra — network automation and generative AI tools
Telecommunications provider Telstra developed a Knowledge Plane that uses knowledge graphs and intent-based reasoning to automate network services. The system processes a higher volume of orders, reduces processing time by 30 per cent and cuts service creation and onboarding times from 12-18 months to just three months. This network innovation has also improved Net Promoter Scores.
Separately, Telstra rolled out two generative AI tools for frontline staff. One Sentence Summary summarises recent customer interactions and saved time for 90 per cent of employees, resulting in 20 per cent fewer follow-up contacts. Ask Telstra allows staff to query internal knowledge bases. Over 80 per cent of pilot participants agreed that the technology improved customer interactions.
Suncorp — generative AI in insurance
Insurer Suncorp is scaling generative AI across operations. Its internal Single View of Claim (SVOC) web application summarises claim details for staff. Early results show that it saves five to 30 minutes per claim review, depending on complexity, and has processed 2.738 billion words, producing about 1.8 million claim summaries. Suncorp is exploring about 120 generative AI use cases, with 20 planned for deployment by mid-2025. These initiatives are supported by a central AI steering committee and partnerships with Microsoft to ensure responsible use.
Journey Beyond — AI in experiential tourism
Journey Beyond, Australia's largest experiential tourism business, adopted AI to support call-centre staff and automate feedback processing. Its ChatJB chatbot provides agents with rapid access to information across 15 brands and helps craft empathetic responses during disruptions. Around 140 employees use the chatbot, with specialised versions supporting attractions such as Melbourne Skydeck. The company also deployed Azure AI Document Intelligence to read handwritten feedback forms. This replaced manual data entry that previously required up to 16 hours per week, allowing feedback to be processed and visualised automatically.
Ashurst — generative AI in legal services
Global law firm Ashurst conducted a controlled trial of generative AI involving more than 400 people across 23 offices. Participants reported an average time saving of 45 per cent when drafting legal briefings, with some tasks seeing up to an 80 per cent reduction in time. These savings equate to about 2.5 hours per draft. The trial found that 61 per cent of respondents felt generative AI effectively reduced workloads, while 88 per cent felt better prepared for the future. A blind study showed that experts misidentified half of AI-generated content, highlighting the quality of AI outputs when human oversight is maintained.
4. AI for new business creation
The above success stories demonstrate how Australian firms use AI to enhance existing operations. Yet AI's transformative potential lies in enabling new products and business models.
AI-enabled product ideation and design. Generative models can accelerate the design of consumer goods by turning market data into prototypes. The design agency Loft uses GPT-4 and image-generation models to brainstorm dozens of novel product ideas, refine features and summarise consumer feedback. This reduces reliance on manual research and shortens design cycles.
Global case studies. L'Oréal integrates generative AI across product development, marketing and virtual try-on assistants. AI reduced product-content development cycles by 60 per cent, rolled out localised descriptions in over 25 languages, and increased user interaction by 35 per cent and conversion rates by 22 per cent.
Biotechnology company Insilico Medicine discovered a drug candidate (INS018_055) using an end-to-end AI platform and advanced it to phase-2 clinical trials in less than 30 months.
Platform company Canva integrated Magic Design and other generative tools into its suite. Within months, users performed over one billion AI-powered actions, with more than 70 per cent of Canva Pro users reporting faster task completion and businesses cutting content-production time by over 60 per cent.
These examples illustrate how AI can become a driver of innovation rather than merely an efficiency tool. They also show that new products often combine AI with domain expertise. Companies that treat AI as a partner for experimentation and exploration are discovering entirely new revenue streams.
5. Recommendations for Australian enterprises
Adopt a whole-of-enterprise strategy led by the board. Productivity and innovation gains materialise when leadership articulates a clear vision, allocates resources and aligns incentives. NBIM's top-down mandate demonstrates the power of unambiguous direction. Boards should set risk appetites, oversee AI strategy and ensure alignment with regulations, including APRA's CPS 230 for financial entities.
Invest in people and processes. Case studies show that the majority of AI investment should go toward training, workflow redesign and change management rather than algorithms. Organisations should create internal AI "enabler" teams and ambassador networks, as NBIM did, to support adoption. CommBank's microlearning series for 43,000 employees and Suncorp's central AI committee are good examples.
Centralise procurement and governance. Shadow use of free AI tools increases risk. Enterprises should centralise the procurement of generative AI solutions, maintain a register of AI systems and require third-party providers to meet privacy and security standards. Telstra's secure partnership with Microsoft and Suncorp's partnership with Microsoft demonstrate the importance of trusted vendors.
Strengthen data governance and privacy. Prohibit staff from entering confidential data into public AI systems. Host models in secure environments and implement real-time monitoring of prompts and outputs. NBIM's private AI platform and Suncorp's Azure-based environments show how to protect sensitive information.
Mandate human oversight and verification. Use AI to augment rather than replace expert judgement. Ensure that decisions, especially those affecting customers or regulatory obligations, remain subject to human review. NBIM forbids AI models from executing trades without human approval. Ashurst's trials show that quality control by qualified professionals remains essential.
Focus on high-impact use cases and measure value. Identify a small number of processes where AI can deliver measurable benefits, then invest heavily in them. Carsales, Telstra, CBA and Suncorp all targeted specific pain points and reported clear metrics such as completion rates, fraud reduction and time savings. Establish baseline performance metrics and track improvements to demonstrate return on investment. Without measurement, AI initiatives risk becoming experiments without impact.
Stimulate innovation and new business creation. Encourage experimentation with generative AI to ideate new products and services. Partnerships with universities or start-ups can bring fresh perspectives and research. Australian enterprises should study global leaders like L'Oréal and Insilico Medicine to understand how AI can create differentiated offerings. The goal is to move beyond efficiency and unlock new revenue streams.
Support SME participation and build national capability. Large enterprises currently dominate AI adoption in Australia. Policymakers and industry bodies should provide training programmes, shared AI infrastructure and grants to ensure that small and mid-sized firms can access AI tools. Expanding adoption will spread productivity gains across the economy.
Conclusion
AI adoption in Australia has moved from experimentation to measurable impact. Success stories across mining, banking, telecommunications, insurance, tourism and legal services show that when organisations invest in people, governance and focused use cases, they can achieve substantial productivity gains and in some cases create entirely new business opportunities. The next challenge is to scale these benefits across the economy while maintaining trust, privacy and human oversight.