With a deep background in machine learning and the ethics of its application, technologist Laurent Giraid has become a leading voice on the integration of Artificial Intelligence within the financial services sector. His work focuses on how legacy institutions can bridge the gap between experimental AI and tangible, bottom-line impact. In this conversation, we explore Barclays’ recent success in leveraging AI for significant cost savings and profit growth. Giraid sheds light on the complex interplay between technology modernization, regulatory hurdles, and investor confidence, offering a blueprint for how traditional firms can operationalize AI to achieve measurable financial targets.
Barclays recently reported a 12% profit increase and set an ambitious new RoTE target, directly linking this success to AI-driven efficiencies. In your experience, what are the most effective ways to actually measure and attribute these kinds of financial gains directly to AI, separating its impact from other market factors?
That’s the crucial challenge, moving from a vague belief in technology to hard proof. While a bank might not publicize the exact internal KPI, the attribution comes from a granular, function-by-function analysis. You don’t just see a 12% profit jump and say, “That was the AI.” Instead, you look at the cost base for specific operations. For instance, in risk analysis, you can measure the reduction in person-hours required to process loan applications or the decrease in false positives for fraud detection after deploying a new model. These micro-level savings, when aggregated across thousands of transactions, ladder up to a significant, defensible figure that can be presented as a driver of overall profitability, distinct from, say, a rise in interest rates.
It’s one thing to run a pilot in a lab, but it’s another to scale AI across a highly regulated institution like a major bank. How does a firm like this successfully navigate the minefield of compliance and data privacy when deploying AI in core functions like customer service or internal reporting?
This is where maturity really shows. A successful institution doesn’t just “bolt on” AI; it builds a governance framework from the ground up. Before a customer service workflow is even touched, legal and compliance teams are in the room defining the guardrails—what data can be used, how decisions are audited, and what the recourse is if the AI makes an error. You see a clear shift from treating AI as a black box to demanding transparency and explainability. For internal reporting, the risks are lower, but the principle is the same: the models are rigorously tested to ensure they don’t introduce biases or misinterpret data from legacy systems. Barclays’ confidence in anchoring its financial forecast on these tools tells me they’ve invested heavily in this internal scaffolding, making compliance a feature of the system, not an afterthought.
When we talk about reducing operating expenses, which for large banks are heavily weighted toward labor and legacy systems, the conversation often turns to job cuts. Could you walk us through a few practical, step-by-step examples of how AI can tangibly reduce costs by streamlining processes, perhaps without a primary goal of headcount reduction?
Absolutely. Think about a process like trade settlement reconciliation. Traditionally, this involves teams of people manually comparing records from multiple systems, a tedious and error-prone task. An AI tool can be introduced first to simply flag discrepancies, making the human teams more efficient. Step two, it learns from their corrections and begins to automate the resolution of common, low-risk mismatches. Step three, it handles the vast majority of reconciliations autonomously, escalating only the complex exceptions. No one has been fired, but the team can now handle a higher volume of trades or be redeployed to more value-added risk analysis. The cost base shrinks because you’ve eliminated overtime, reduced operational losses from errors, and avoided hiring more staff as business grows. The savings are real and measurable, embedded directly in the operational workflow.
When a bank’s leadership stands up and connects a plan to return over £15 billion to shareholders with its AI strategy, that’s a bold statement. How do you effectively communicate this to build investor confidence and prove that these AI investments are delivering real value, not just chasing a trend?
The key is to ground the narrative in concrete financial outcomes. It’s not about showcasing a futuristic AI agent; it’s about saying, “Our investment in technology platform X has reduced our cost-to-income ratio by Y percent this quarter, contributing Z million to the bottom line.” Barclays did this brilliantly by announcing its 12% profit increase in the same breath as its technology-driven cost-cutting. This creates a powerful, direct causal link in the minds of investors. You present AI not as a speculative R&D expense, but as a core pillar of your cost discipline and operational leverage. The £15 billion return becomes the believable outcome of a strategy that is already delivering proven, quantifiable efficiencies today.
The strategy you’re describing involves weaving AI into broader initiatives like cost discipline and modernizing legacy systems. How critical is that interdependence? Can you give an example of how a promising new AI tool might completely fail if the underlying tech stack isn’t updated in parallel?
That interdependence isn’t just critical; it’s everything. An AI tool is only as good as the data it can access and the systems it can influence. Imagine a bank develops a state-of-the-art AI model for dynamic credit risk assessment. But, its customer data is fragmented across a dozen old, siloed mainframes that can only communicate through slow, nightly batch processes. The AI model needs real-time data to be effective, but the legacy infrastructure can’t provide it. The project would be a catastrophic failure—not because the AI was flawed, but because it was starved of the data it needed to function. This is why Barclays’ approach of trimming the legacy stack while investing in AI is so smart. Modernizing the foundation is what allows the powerful new tools to actually do their job and deliver on their promise.
Do you have any advice for leaders at other legacy firms looking to replicate this success and use AI to drive measurable financial performance?
My advice is to stop treating AI as a separate, futuristic project and start integrating it into your core financial and operational planning. Don’t ask, “What cool things can we do with AI?” Instead, ask, “What are our biggest cost drivers or revenue challenges, and how can AI be a tool to solve them?” Frame every investment in terms of its expected impact on a key metric like RoTE or operating margin. This forces a discipline that connects technology directly to business value. It also means empowering a cross-functional team of technology, finance, and compliance experts to work together from day one. True transformation happens when AI is no longer the responsibility of just the innovation lab, but a fundamental lever in the CFO’s toolkit.
