As Flutter Entertainment moves into a new chapter as a New York–listed company, its finance team is taking a hard look at how artificial intelligence and automation can help manage the complexity that comes with being a global gaming powerhouse.
Best known to punters through brands like Paddy Power, Betfair, Sky Bet and PokerStars, Flutter has grown rapidly through acquisition. FanDuel leads its operations in the US, while Sportsbet dominates in Australia.
Each brand retains a degree of autonomy, especially in front-end operations, but for back-office teams like finance, integration is essential.
James Smith, finance automation lead at Flutter, is part of a team tasked with turning that patchwork into a coherent system. “Where we have finance processes that are consistent across all these brands, we’re trying to consolidate, automate and control them,” he says.
The challenge is both operational and regulatory. Last year, Flutter shifted its primary listing to the New York Stock Exchange, subjecting it to the Sarbanes-Oxley Act (SOX).
That move has added new layers of scrutiny to financial reporting, pushing the company to eliminate manual processes and improve visibility across disparate systems.
Rebuilding the finance backbone
Smith joined Flutter five years ago in what was then the Finance Business Intelligence team. Its job was to translate transactional data, such as settled bets and free bet usage, from data warehouses into meaningful accounting metrics. But with different divisions using different systems, accessing the right data was cumbersome.
“Analysts used to have to switch onto three or four different VPNs to access data – from FanDuel, or Sportsbet, for example,” Smith explains. “We did a lot of hard work to get our server connected to all these different places so we could query them from a central point.”
That groundwork underpins the group’s evolving finance automation stack, initially built around tools from data and AI firm Alteryx.
While Flutter has since expanded its toolkit, Alteryx remains a central piece of the puzzle, allowing teams to directly connect to enterprise resource planning (ERP) systems like Oracle, perform data preparation, and cut out repetitive Excel-based steps.
“Previously, people would download standard Excel reports from Oracle, process them manually, and maybe create a journal to go back into the system,” Smith says. “Now, we’re much closer to the point where someone is reviewing information, rather than preparing it.”
AI as a copilot, not an autopilot
Flutter’s finance team is still in the early phases of deploying AI, particularly when compared to customer-facing areas like product and support. However, Smith sees potential – particularly in embedding AI into existing tools and workflows rather than standing up standalone models.
One area of focus is agentic AI, where autonomous systems can fetch, compare and validate data across platforms.
A common scenario is invoice matching: identifying discrepancies between a purchase order and a submitted invoice. For now, these tasks are handled manually – but not for much longer.
“An agent could go and get that data from different places, do the comparison. 95% of the time, it would get it right. And in the 5% of cases it doesn’t, you could review the steps it took and see where it went wrong,” he says.
Notably, Smith isn’t advocating for AI to take full control. Instead, he sees it as a way to bring humans closer to the final decision point while eliminating the drudgery of data prep and reconciliation.
“It’s not just input to output. You can see what the AI did at each step, like an analyst would. That’s how you build trust in the output,” he says.
Internal LLMs and building trust
While generative AI tools like ChatGPT have captured public attention, Flutter has taken a more tailored approach. The company has built its own large language model (LLM) on internal documentation.
To build confidence in automation, the finance team has adopted a “parallel run” strategy, running AI-enhanced processes alongside existing ones, validating outputs, and ensuring consistency before switching over.
“That’s how we’ve built trust with end users,” James says. “There’s also a lot of governance and people checking what we’re doing.”
That governance is becoming more important than ever. The shift to SOX compliance means every automated step must be auditable and transparent. But rather than seeing this as a hindrance, James views it as an opportunity.
“SOX has been a burden – but it’s also driving better processes. We’re using this as a chance to bring in automation and unlock faster reporting.”
A bet on future gains
Flutter is not alone in this journey. At a recent panel on enterprise AI adoption, James shared the stage with data leads from Sainsbury’s, EDF and the NHS Business Services Authority. Each spoke about balancing innovation with compliance, using automation to relieve staff of repetitive tasks, and the need for solid data infrastructure before AI can deliver real value.
Smith’s advice to others considering AI in finance? Start small, and show tangible results.
“If you can’t do it at work, experiment with your own data,” he said. “Show a use case, then build the case for adoption.”
