As AI moves from experimentation to enterprise adoption, finance leaders are being challenged to distinguish meaningful innovation from market noise. That was the focus of Beyond the Hype: Practical AI for CFOs, where Allura Partners brought together senior finance executives for a grounded, insight-led discussion on what AI is truly delivering today. The central question: how can CFOs move beyond the AI hype to achieve genuine business impact?

Hosted by Allura Partners, the timing for this interactive event was spot on. According to recent McKinsey research cited during the evening, only about 6% of companies qualify as 'AI high performers', organisations that achieve a significant enterprise-level financial impact of 5% or more EBIT from AI initiatives. Most companies are still struggling to scale AI beyond localised use cases to achieve enterprise-wide financial returns. Yet the technology is advancing at breakneck speed, the potential to drive efficiencies and growth is enormous, and the window to build competitive advantage is rapidly narrowing.

Why Finance Leaders Must Act Now

The evening’s speakers were Laetitia Andrac and Johan Erchoff. Laetitia, who led digital expansion in various roles with Telstra from 2014 to 2021, worked with AI “before AI was cool”. Her experience navigating enterprise-scale transformation at one of Australia’s largest organisations provided practical grounding for the evening’s insights.

Johan, with his background in strategy consulting and experience launching subscription businesses from the ground up, complemented Laetitia’s technical depth with commercial pragmatism. Together, as co-founders of AI consultancy Louvea and social enterprise Understanding Zoe, they’ve witnessed firsthand how organisations struggle – and can succeed – with AI adoption.

Their central message did not go unheard: intelligence has become a commodity. What once required teams of analysts or data scientists is now accessible through simple tools. For finance professionals, this democratisation of cognitive work represents both unprecedented opportunity and existential challenge.

The Business Case for AI in Finance

The numbers presented during the evening painted a strong argument for AI adoption. Those organisations effectively deploying AI are seeing three times the growth per employee – not simply because individuals are working harder, but because AI accelerates time-to-market, enables faster experimentation, and unlocks capacity for higher-value work.

Goldman Sachs projects that generative AI could drive a 7% increase in global GDP, or almost $7 trillion, over a 10-year period. For finance functions specifically, 72% of organisations globally have already integrated AI into their processes in some form, according to a KPMG study of 2,900 companies across 23 countries. 

And the benefits extend beyond the balance sheet. Salesforce research shows 92% of decision-makers say generative AI helps them deliver better customer service, whilst Deloitte found that 82% of surveyed leaders state AI increases job satisfaction and enhances performance.  Rather than destroying jobs, AI eliminates tedious, repetitive tasks, freeing finance professionals to focus on strategic analysis and business partnerships.

Laetitia, who disclosed to the audience that she is autistic, emphasised AI’s democratising potential: “For neurodivergent individuals, people with disabilities who were excluded from the workforce, AI is levelling the playing field. If you’re dyslexic, you can use AI to draft emails. It’s making the world more inclusive.” 

Three Levels of AI Maturity

• Basic Level: Task Automation
  Most organisations start here – using AI for basic spreadsheet work, or simple prompts to ChatGPT or similar tools. The benefits include time savings, reduced cognitive overload, fewer errors, and immediate productivity uplift.
• Intermediate Level: Predictive Decision-Making
  At this stage, AI begins learning from patterns and predicting outcomes, though always with a human in the loop (as Laetitia observed, research shows organisations maintaining human oversight achieve three times better results than those attempting full automation prematurely).
  One attendee shared how their team uses AI for variance analysis. “Instead of digging into every single movement (with AI), I've already identified the top 80% of variants that need my attention. It accelerates what I already do,” they said.
• Advanced Level: Agentic AI
  This frontier involves autonomous agents that don’t merely answer questions but take actions – reconciling invoices, updating systems, and generating compliance reports. Multiple agents can work collaboratively with human teams, handling repetitive work while leaving judgment, creativity, and relationship management to people.

Currently, only 6% of companies qualify as 'AI high performers' (Understanding Zoe being one of them) operate at this level, but the speakers argued that the tools and capabilities now exist to make this accessible to mid-sized organisations, not just enterprise giants.

The Reality Check: Starting Small

Several questions from the floor revealed common barriers. One CFO asked about organisational stigma – colleagues viewing AI use as “cheating” rather than efficiency. Another questioned investment requirements, referencing cases of organisations spending hundreds of millions on failed AI programmes.

The speakers pushed back against both concerns. On stigma, Laetitia advised, “Prove them wrong through action. Show how AI improves quality, saves time, and enables your team to focus on high-value work. You need to become the AI champion in your organisation”. 

Regarding costs, they argued the landscape has fundamentally shifted. “Last year, agentic AI was expensive and complex. Now, token costs have dropped dramatically. You can achieve significant automation with modest investment if you start small and iterate.”

While acknowledging that an expert consultant is an essential partner when implementing AI, they recommended this five-step approach to get started:

1. Identify one genuinely annoying, repetitive task,
2. Experiment with automating it using AI,
3. Measure the results rigorously,
4. Scale what works; kill what doesn’t, then
5. Build organisational muscle through repeated cycles.

Five Principles for Success

Additionally, drawing on their professional experience, Laetitia and Johan offered a five-point framework:

• Business First, AI Second: Start with strategy and desired outcomes, then determine which AI tools fit the need – not the reverse.
• Bottom-Up Energy: Empower teams to experiment; 78% of workers already use AI without approval, so creating safe, sanctioned pathways channels this energy productively.
• Flexible Strategy: Have a plan, but adapt rapidly as capabilities evolve.
• Clean Foundation: AI trained on poor-quality data produces poor-quality insights. Invest in data hygiene.
• Iterate to Scale: Move through quick experimentation cycles, learn from each attempt, and be ready to expand successful pilots into full-scale operations rather than leaving them as perpetual proofs-of-concept.

Managing the Risks

While the audience was encouraged to get started sooner rather than later, Laetitia and Johan also called out five critical risks requiring active management:

• Hallucination: AI confidently generating incorrect information remains problematic. The speakers recommended always fact-checking outputs and even deploying secondary AI agents to verify results. As Johan noted, “The more you tell AI not to hallucinate and to focus only on verified information, the better it performs”.
• Privacy: With finance teams handling sensitive data, robust governance frameworks are essential. Simple measures, like disabling data feeds in ChatGPT, provide a starting point, but organisations need clear policies about what information can be processed through AI systems.
• Sustainability: AI models consume significant energy. Thoughtful deployment means considering which AI model is most suited to the project and even which tasks genuinely require AI processing versus where traditional methods suffice.
• Creativity: The evening began with a demonstration – asking attendees to prompt AI for a random number between one and ten. Most received seven, revealing how AI learns from patterns rather than truly generating randomness. “Don’t let AI think for you,” Laetitia warned. “Keep your human creativity”.
• Transparency: Both speakers advocated for disclosure when AI is used, maintaining trust in an era of deepfakes and synthetic content.

The Cost of Waiting

Perhaps the evening’s most sobering insight: McKinsey research suggests organisations failing to adopt AI by 2027 will face significant competitive disadvantage. 

As one attendee noted while describing their AI-assisted workflow, the technology doesn’t replace the expertise – it removes the first 60% of grunt work, letting professionals focus on the interpretation, judgement, and strategic insight that can really differentiate performance.

For finance leaders, this isn’t about chasing technology trends – it’s about maintaining relevance as the tools of the profession fundamentally transform.