How LIV Golf is using Agentforce and Data 360 to bring fans closer to every swing.

AI doesn’t work without a lot of organized, diverse data, which is why, when I stepped into my role, building a strong data foundation was my number one priority. My first six months were really about building a platform we could safely accelerate on. The goal was to have a unified data foundation that would allow us to fully understand who our fans are, what they’re interested in, and when and how they engage with us — and to have that foundation in place before the League’s 2026 season.

To get there, we needed a trusted, complete view of our fans across all the different ways they interact with LIV Golf. I knew Data 360 was going to unlock that for us. It allows us to unify fan, player, sponsor, and operational data across the business.

What excites me now is what comes next. Data 360 allows us to act on our data – scaling use cases across AI agents, marketing segmentation and merch sales, accelerating interactions, and improving experiences across the world.

Data is created at nearly every touchpoint. Like with events, whenever someone scans their ticket, taps an RFID tag entering hospitality, or swipes a credit card for food or merch — that’s data. Digitally, fans create data when they stream, read articles, check stats, or engage with loyalty, gamification, and fantasy league platforms. Of course, there’s also transactional data from ticketing and retail.

We also have data on the sport itself, including how our athletes are performing and what clubs they’re using. It’s an especially exciting time because we’ve moved to a 72-hole format. This format means we can capture more shots, swings, clutch moments, and micro-stories within tournaments we’re already producing. That’s a lot more data, and for us, that’s a great opportunity to have, because the structure remains the same.

Data 360 connects all of those signals into a single profile in near-real time, it allows contextual engagement — the right message, in the right channel, at the right moment. That’s where Agentforce comes in, creating true two-way conversations with fans in a way that simply wasn’t possible before.

Two moments really stand out. The first was when Agentforce was announced at a Salesforce event two years ago. Seeing demonstrations around Prompt Builder and topic creation immediately sparked ideas. The ideas started exploding – I could see Agentforce being deployed in so many ways both for fans and across our enterprise.

The second moment came about six months ago, when I realized something that changed the whole game for me: Fans don’t think in dropdown menus — they think in questions. That’s the moment I could really picture the value of using Agentforce to answer questions conversationally and instantly, without asking fans to search or click through menus.

It really came down to asking two questions: What would bring the most value to our fans, and what would have the highest adoption? We felt strongly that our two first AI agents – Fan Caddie and Agent Caddie – would materially improve experiences for our viewers and broadcasters.

Fan Caddie, which we’ve named Chip, is focused on fans. It delivers news, recaps, and performance stats on players and teams throughout the season. Fans can be watching an event — in person or on screen — and engage with Agentforce on their phone for that second-screen experience. It allows them to really understand exactly what’s happening, why it matters, and how it fits into the bigger story.

Agent Caddie is built for broadcasters. It surfaces context and insights that enable real-time, data-driven storytelling on air.

We also thought a lot about adoption — not just how we would engage with fans, but what they would want to engage with in return. That idea of a two-way conversation was really important to us.

These use cases bring value to LIV Golf as well. Interactions with Fan Caddie help build stickiness with our products and greater affinity with our brand, teams, and players. It doesn’t matter if a fan is watching 10 feet from the fairway or from a continent away through one of our broadcast partners or a LIV Golf platform. Everyone can have a second-screen experience, asking about stats, leaders, birdies, courses, or even the hat Bryson DeChambeau is wearing and purchasing it through Fan Caddie. Agentforce makes those interactions really come to life.

The build has been interesting. We turned the first prototype around in about three days. Building an agent with Agentforce is simple — but getting the data right to ensure Agentforce responds accurately and consistently wasn’t without its challenges. We learned:

1. Data quality is essential. You have to make sure your data is organized and accessible to Agentforce, because otherwise you’re going to get poor answers. I don’t think there’s a clear playbook yet for how that data should be structured. We went through a few iterations before landing on the best way to organize and expose data, which was particularly important for Fan Caddie, because users are asking for statistical information.
2. Pay attention to the question type. Knowledge-based questions are fairly black and white. For example, when asked, “What are my options for watching the event in the UK?” Agentforce consistently provided a straightforward answer about the different broadcasts available. But statistical questions required more work. Those questions can be interpreted more broadly, which creates risk for inconsistent answers and unreliability if the data isn’t extremely well defined. If a fan asks how many birdies a player has, Agentforce needs a clear way to distinguish whether they mean birdies in a specific tournament, across the season, or an average rate per hole.
3. Metadata matters. Our challenge was partly related to how the data was engineered and partly to how it was retrieved. We have a metadata layer that includes every field, along with a clear definition of what each data point represents. Agentforce needs a very precise understanding of those definitions in order to return accurate, consistent answers. Once we established a strong, repeatable pattern for organizing our metadata, the results improved significantly — and that pattern will be the foundation for future use cases as we continue building our Agentic Enterprise.

I like to be very close to new rollouts, so I personally ran a 200-question round of QA and shared Agentforce’s answers with Salesforce Professional Services. We noticed that some questions should have been answered by Agentforce — it had access to the right, well-organized data — yet the agent was still saying, “I don’t know.” We discovered that because Agentforce was configured with multiple topics, it was sometimes routing questions to the wrong one, which caused it to get confused and either provide incorrect answers or not answer at all.

Within three days, Professional Services consolidated and rewrote the topics — which really speaks to the speed to market here — and I tested it again. The answers were significantly better than what we were getting before. I think that shows the power of Salesforce and Agentforce. You have full control, and when you update the prompts within your topics, it has a tangible impact on answer quality. Ultimately, that means you can trust that Agentforce’s answers are the ones you expect.

We’re a young company, so we should be AI-native to help us continue to innovate and adapt. We have over 40 Agentforce use cases in mind, including internal agents for IT support and advanced marketing segmentation activated through Agentforce Marketing.

One aspiration I have is helping people access our events globally. Imagine someone saying, “I live in California and want to go to the Andalucía event,” and Agentforce builds their itinerary — flights, hotels, ticket types — and they just click “book.” That’s where I’d love to go.

Focus on your data first. AI isn’t deterministic like a SQL query, and not all tools are created equal. You need to ensure the quality of your underlying data is properly available to your agent.

It’s worth the setup. The Agentforce 360 Platform effect — Agentforce connected to Data 360 and Agentforce Marketing — gives you a centralized customer view you can activate everywhere. Building with Agentforce isn’t hard. Focus on getting the data architecture right and it’ll work brilliantly.