The Realities of AI
AI adoption in the industry is being guided by strategic use cases and cautious optimization. Respondents rank fraud detection (7.47), algorithmic trading (6.78), and customer service chatbots (6.3) as the most applicable AI use cases - indicating a focus on efficiency, risk mitigation, and client engagement.
Despite this, many firms are still refining their expectations. Forty per cent said the impact of AI initiatives delivered on their original promise, but 33% experienced higher-than-expected costs, suggesting persistent challenges in cost control and implementation. It is clear that running modern AI infrastructures is not cheap and any gains in productivity are potentially lost to infrastructure costs in 3 out of 5 options.
For underperforming AI projects, firms are adopting more disciplined and collaborative approaches. There is a clear pivot towards explainability, risk analysis, and model transparency, often at the expense of complexity. Regular retraining cycles with new data, structured validation, and broader stakeholder involvement have become common practice. Firms are embedding cross-functional collaboration early in the development process - pairing data scientists with business leads and compliance teams to ensure alignment with operational realities.
Strategically, firms are also institutionalizing AI oversight with dedicated governance structures, centralized repositories for past projects, performance review protocols, and agile playbooks. These measures reflect a maturing AI landscape where business value, control, and responsible innovation are being balanced more deliberately, which is laying the groundwork for more scalable, reliable, and accountable AI deployments.
"AI adoption in the financial industry is increasingly driven by strategic, high-impact use cases such as reconciliations, fraud detection, client personalization, and regulatory compliance, rather than broad, unfocused implementation. Reconciliations, in particular, are a prime target for automation because they reduce manual effort, improve data accuracy, and speed up financial close cycles. Institutions are approaching deployment with measured caution, focusing on optimizing existing processes, ensuring explainability, and mitigating operational and reputational risks. This combination of targeted innovation allows firms to capture value while maintaining regulatory trust and operational stability."
Julian Trostinsky, Global Director of Solutions Engineering, Gresham
Question 1: Rank the following AI use cases in order of most applicability to your firm
(1= least applicable, 9= most applicable. Figures below represent the average rank)
Fraud detection and prevention
Algorithmic trading
Customer service chatbots
Credit risk assessment
Personalized financial advice
Regulatory compliance
Process automation (RPA with AI)
KYC/AML (Know your customer/Anti-Money Laundering) processes
Market sentiment analysis
"It really depends on the role within the organization — artificial intelligence can be applied across a vast number of areas. But definitions vary. Some people think of AI as just a chatbot interface, like generative AI trained for client interactions, while others include data science, machine learning, or advanced automation under that umbrella.
From our side, we’re seeing promising results in the operations space, particularly around reconciliation, matching, and settlement. There's enough data in these areas to effectively train algorithms to identify patterns and streamline tasks. We're also observing use cases in reviewing contracts and documentation, thanks to advances in deep language models. AI is playing a growing role in fraud detection and even marketing content generation. Ultimately, I think every organization will find its own use cases for AI, depending on their structure and needs — and we’re already seeing that start to take shape across the industry."
Krzysztof Wierzchowski, SVP Business Transformation, Franklin Templeton
Question 2: How has the actual business impact of your AI initiatives differed from initial expectations? (e.g., cost savings, efficiency gains)?
It is as expected
It is slightly more expensive than expected
It is slightly cheaper than expected
It is much more expensive than expected
"I’ve seen it with every new technology — it starts off expensive, but over time, as experience builds and more use cases emerge, it becomes more affordable. That’s exactly what we’re seeing with artificial intelligence. The time it takes for new tech to reach maturity has compressed significantly in recent years. What once took years now happens in a much shorter cycle, and AI is a clear example of that trend.
My expectation is that as costs continue to fall and adoption increases, the use of AI will only grow. The impact on organizations will become more substantial, especially if we can continue to identify and prove out high-value use cases. Once it’s clear that AI can deliver at scale, it becomes a very effective way to reduce the cost of operations — and that’s where I think the real value will come through."
Krzysztof Wierzchowski, SVP Business Transformation, Franklin Templeton
“From my perspective, leading the group, I expect AI will take on an increasingly important role across the industry. It has the potential to address several long-standing deficiencies that we've struggled with, particularly in areas that are traditionally labour-intensive and reliant on manual processes.
This is clearly a move in the right direction. Rather than spending time trying to investigate why transactions are breaking or mismatches are occurring, we want to redirect our effort toward areas that generate more value. AI is very well suited to support us in that shift — increasing both effectiveness and efficiency where it matters most.”
Krzysztof Wierzchowski, SVP Business Transformation, Franklin Templeton
