How the Premier League with Microsoft AI Guides FPL Managers

Carrying the football World Cup momentum into UK's domestic games, the Premier League and Microsoft are using gen AI to transform Fantasy Premier League (FPL) for more than 13 million managers. 

FPL is the official fantasy football game of the Premier League, where participants act as virtual managers with a £100m (US$133m) budget to assemble a 15-player squad of real-world athletes.

Users select a starting 11 each matchweek, earning points based on real-world player performances through goals, assists, clean sheets and bonus metrics.

To help managers navigate these choices, the organisation is launching the Fantasy Premier League Companion, a new conversational AI feature designed to act as an intelligent sounding board for squad decisions.

Built on Microsoft technology, the Fantasy Premier League Companion uses Microsoft Foundry, Azure OpenAI and gen AI technology based on ChatGPT 5.4. 

The conversational interface brings together official Premier League match data and FPL game metrics to help users construct and maintain their squads.

Darren Hardman , CEO of Microsoft UK & Ireland, says: “The Fantasy Premier League Companion, powered by Microsoft Copilot, is designed to help managers make more informed decisions every week, bringing together official Premier League data, personalised team context and conversational AI to make the game easier to understand and more engaging to play.

“Whether you're new to FPL and need help understanding the rules of the game, formations or scoring or you're an experienced manager looking to compare strikers by shots in the box, fixtures or form, the Companion is built to provide useful insights quickly and clearly.”

Fantasy gaming with official match data

Building on the foundation of the existing Premier League Companion, the FPL Companion provides an intuitive gateway for both newcomers and experienced fantasy participants. 

The tool brings a wealth of data into a conversational format that is easy to understand. The underlying technology processes two distinct datasets to deliver tailored insights:

• Match data, encompassing goals, assists, appearances, cards, minutes played, tackles and shots in the box
• FPL game data, including transfer trends, captaincy choices and chip usage.

Instead of requiring managers to navigate complex tables, the tool combines these streams to answer specific player queries, such as identifying potential replacements based on recent form and upcoming fixtures.

Despite its depth of analytical capability, the FPL Companion is designed to support human managers rather than run squads automatically. 

Andy Selby, Senior Gaming Manager at the Premier League, emphasises that the system will not predict fantasy points or act as an autopilot, keeping decision-making power firmly in the hands of the user.

He says: “It is definitely not there to be an autopilot for your team. It will not predict FPL points. It is very important to us that FPL managers themselves are making their own decisions. You are the manager of your own team; it is your call at the end of the day. The role of FPL Companion is not to do that instead of you. It is to help you make better, informed decisions.

"Because there's never one single right answer. Football by its very nature is very unpredictable. So, if I had someone on my team and he's injured, Companion might say, okay, you could look at [Chelsea's] Cole Palmer, or [Arsenal's] Bukayo Saka or [Manchester City's] Antoine Semenyo." 

“It will always give you a number of options. It will give you the relevant data to back it up. It will link to a relevant editorial where we've tackled similar questions but you then as the manager are going to go and make that decision.”

Guiding new managers and strategic experts

The conversational interface is engineered to adapt dynamically as a user’s technical understanding of FPL evolves over the campaign.

For new managers who find initial squad selection daunting, the system provides answers to fundamental queries like how to set up mini-leagues with friends, explaining how FPL chips operate and recommending foundational team composition tactics.

As managers gain experience, Andy notes that the tool moves alongside them, catering to advanced statistical queries. 

Experienced users can interrogate the system about specific player metrics, such as identifying goalkeepers with favourable upcoming fixture runs or strikers taking the highest number of shots inside the penalty box.

The platform also reflects shifting fan behaviours, where supporters increasingly track individual athletes across the sport alongside traditional club loyalty. 

By updating in real time as press updates and squad news break, the tool keeps fans engaged throughout the matchweek. A query submitted on Tuesday may return updated recommendations by Thursday if new injury reports or team developments occur.

The Premier League plans to iterate the tool across the season by analyzing user interaction, response patterns and follow-up questions. 

This real-time feedback loop will allow developers to refine answer accuracy and continually optimize how data is surfaced.

Andy adds: “We will listen to users. We'll be putting this out into the wild ahead of the new season and very quickly we'll be able to see the sorts of questions that users are asking. Where are they spending the most time with this? What are the follow-up questions that are being asked?”

As managers engage with the platform throughout the campaign, the tool will iteratively evolve to enhance its conversational capabilities and surface increasingly relevant insights.