Wimbledon's Match Chat answers fans live while editors keep human oversight
14M+According to SportsPro: Wimbledon's earlier AI features generated over 14 million views of pages that contained AI-powered output, and about 27 million unique users engaged across its digital channels over the full year. No usage figures for Match Chat itself were published.
“This has resulted in greater engagement across Wimbledon’s platforms, generating over 14 million views of pages that contained AI-powered output. The Grand Slam has also seen about 27 million unique users engaged for the full annual year across its digital channels.”
SportsPro, Jun 2025 (also Sports Business Journal, Jun 2025), 17 Jun 2025
- Who
- The All England Lawn Tennis Club (AELTC), organiser of Wimbledon, and its official technology partner IBM (partners for more than three decades). Match Chat built on IBM watsonx Orchestrate, IBM's AI agent platform, with IBM Granite large language models and watsonx.governance.
- Problem
- Fans following a live singles match want quick answers (who has converted more break points, who is playing better) without digging through stats pages. The 2024 AI feature Catch Me Up had drawn criticism from some users for factual errors in player profiles, so any new AI feature needed tighter control.
- Agent
- Match Chat is a conversational assistant in the Wimbledon app and on Wimbledon.com. Fans tap a pre-written prompt or type their own question about a live singles match; it answers immediately from match insights. It runs on IBM watsonx Orchestrate with Granite models trained on Wimbledon's editorial style and tennis terminology, and watsonx.governance acts as a control layer for accuracy and style.
We'll build it for you
Book a callView the prompt
Try it yourself with IBM watsonx Orchestrate
You help run a tennis or padel tournament or a club league. Goal: a live Q&A assistant that answers fans' questions about a match from our own data, reviewed by my team before launch. My input: [live score and stats feed or a sample export], [player profiles we have approved], [our editorial tone and words we use], [questions fans asked us last season]. Give me: 1) the 15 most likely fan questions and the stat each one needs; 2) a short answer template per question using only my data; 3) rules for when to say 'I don't know' instead of guessing; 4) a checklist my team uses to approve answers before going live and to spot errors during the event. Never invent stats or quotes.What we can do for your club or brand
Most tournaments and clubs field the same live questions every match day: scores, stats, who is playing next. The Court can set up an AI assistant that answers fans and parents from your own results and player data, in your tone, with your team approving the answers and watching for errors, so your staff spend match day on the event and you learn what your audience actually asks.
What they did
For the 2025 Championships, starting 30 June, the AELTC and IBM added Match Chat to every men's and women's singles match in the app and on the website. The 2024 Catch Me Up player analysis was folded into it. IBM said watsonx.governance controls accuracy and alignment with Wimbledon's editorial style, plus human oversight in the preparation and monitoring process. IBM also planned to measure fan engagement before considering expanding it to Wimbledon's other events.
How it went
SportsPro reports that Wimbledon's AI features had generated over 14 million views of pages that contained AI-powered output, with about 27 million unique users across its digital channels over the year. Chris Clements, digital products lead at the AELTC, said Match Chat "will also give us some really valuable data around what the most common questions are that we can use in the future", and that the human element stays "so vital, especially when you're talking about sport." Limits: the 14 million views and 27 million users describe Wimbledon's AI features and digital channels overall, not Match Chat alone; no Match Chat usage or accuracy figures were published; the 2024 Catch Me Up feature was criticised for factual errors, hence the human oversight.