With the hype cycle over, the attention around artificial intelligence (AI) is now about value creation. Executive teams, investors, boards, and more are asking when the promised gains will hit the bottom line. This means that the teams applying AI to products, processes, and more must quickly move from the experimentation or proof-of-concept (POC) state, to scale.
This challenge is what united attendees at Beyond PoC: Accelerating Impact from AI, a summit hosted by QuantumBlack, McKinsey’s specialist AI arm, in Sydney. What emerged is the important role humans play in the successful scaling of AI; the need for direct conversations and clear commitment across the business; and the humbling yet empowering acknowledgement that everyone is still grasping the true power of this transformative technology.
Moving pilots from excitement to impact
AI pilots have proliferated over the past two years, but only 10–20 percent of isolated AI experiments scale to create value. This can lead to fragmented efforts, where AI projects exist in silos. What is required is a domain-level approach, where entire divisions are reimagined and transformed with AI. This also means assets can be applied or adapted to other parts of the organisation to not only increase the return-on-investment, but adoption and business integration.
Shifting from technology- to outcome-centric approaches
Amid hype, excitement naturally surrounds the technology and its capability. But to achieve value through scale organisations must focus on the broader ecosystem of people, processes, and operating models. It’s “never just about the tech” or implementing the latest algorithms and models, but coupling the technology with people around it.
One example shared was an Asia-Pacific bank that is leveraging gen AI within mortgage operations to help capture up to $100 million of impact. By assisting routine mortgage assessments, the solution will help accelerate the time to decision, and enable human assessors to focus on more valuable, complex tasks. This will not only improve operational efficiency but also deliver better outcomes for customers.
Another South East Asian bank delivered an organisation-wide transformation by: accelerating value from vertical use cases like developer and contact centre productivity; driving adoption through horizontal gen AI capabilities, like summarisation, generative communications, and consolidation of common knowledge; and industrialising its delivery engine across use cases, like creating a resuable set of common components to accelerate use case development. More than four in every five employees are now enabled through gen AI, and the organisation is well on its way to a goal of a 15%+ productivity uplift.
Lessons from successful scalers
Several practical strategies for those looking to “go beyond POC” emerged over the day.
1. Break out of ‘pilot purgatory’ by focusing on domains
Successful companies concentrate effort on a few key domains – be it finance, or marketing, or a product line – rather than spread resources thin across multiple pilots. This not only drives synergies across data, technology, and people, but also lowers the cost per use case, delivering more substantial impact. One standout example from the day came from a leading retailer that will integrate generative AI into its merchandising and design process, allowing it to better and more quickly understand and respond to what products perform well in-market.
2. Prioritise end-user adoption
Successful scalers invest heavily in training and capability building – which can often require as much, if not more, effort than the actual technology implementation. One company saw a threefold increase in user adoption after just one hour of hands-on training. Another customer-focused example was a large life insurer that achieved a 10% cost reduction and a $250 million impact by accelerating its digital transformation. By becoming a truly customer-focused, data-driven organisation and improving its organisational health, the insurer was able to streamline operations and significantly enhance customer experience and cost efficiency.
3. Transform operating models for sustained value
This requires optimising workflows, customer journeys, and personas to continuously capture and reinvest AI-driven efficiencies. The whole leadership team should be involved in shaping the AI agenda, creating a shared ambition that aligns with broader business goals.
4. Build modular, scalable AI infrastructure
Scaling is significantly easier with reusable technical components and playbooks. It enables faster and safer scaling across multiple use cases, reducing the cost and complexity of AI implementations. By focusing on scalability via this modular approach from the outset, organisations can replicate successes and achieve broader impact more efficiently.
5. Balancing short- and long-term value capture
Successful companies create incremental improvements for immediate impact, while maintaining a clear vision for transformative change that creates the space to capitalise on AI’s evolving potential.
Unlocking the full potential of AI – and the country
Scaling AI is a technical and organisational challenge that many are facing. But as Jonathan Michael, Senior Partner and Leader of McKinsey Digital across Australia and New Zealand put it, what is exciting is how the leaders gathered were committed to not only pushing their organisations forward, but in doing so, the entire country.