How TWE is Transforming the Wine Industry with GenAI

00:09)

Hello, everyone. I'm Nicola Quail, one of the co-founders of the AI Marketers Guild APAC, along with my colleagues De Shama Badi Shama from India and Sushita Mahapatra from Singapore. We're excited to have you all here for what we hope will be one of our biggest AI Marketers Guild webinars to date. It's great to see this community expanding across APAC, and we even have some colleagues dialing in from North America.

Before I introduce our two incredible guest speakers, I want to give a quick background for those new to the community. The AI Marketers Guild was the brainchild of US marketer David Burwitz. And for those familiar with his LinkedIn, yes, that David Burwitz.

After launching two years ago, the Guild has grown significantly in North America, especially given AI’s impact on marketing. We reached out to David to partner with AIMG and create a regional community to showcase local AI pioneers, tools, applications, and best practices while integrating world-class AI thought leadership.

We've been fortunate to have so many professionals join us over the past six months. We host monthly webinars like this one and will invite you to join our Slack community afterward. We're also hoping to organize city-based meetups later this year.

Rather than just observing from the sidelines, we love that you're all actively engaging.

(01:45)
On that note, I’m thrilled to introduce two marketing pioneers today. I must thank our mutual colleague and friend, Adam Murphy, for introducing me to Adele and Maddie from Treasury Wine Estates.

Managing a large global marketing insights function is challenging enough, but how do you streamline and create efficiencies in managing and distributing all the data and insights within a complex business? This was one of the many challenges Adele and Maddie set out to solve using AI.

While I admire the technical aspects of what they've achieved, what truly inspires me is their entrepreneurial spirit in tackling this problem.

Without further ado, I’d like to hand it over to Adele Fou and Maddie Simik to share their story. They’ll present for about 25–30 minutes, and then we’ll have time for questions. Please drop your questions into the Q&A or chat, and we’ll address them at the end.

Welcome, Adele and Maddie!

(02:47)
Adele: Awesome, thanks, Nicola, and thanks, everyone, for having us. First time being called a marketing pioneer—I’ll take it!

I’m not actually a marketer by background, but I’m embracing it and starting to love the world of marketing. Maddie and I are excited to be here, though this is a bit outside our comfort zone. We’re not usually the ones to step up and share about ourselves or our work, so bear with us!

That said, we’re passionate about this project and excited to showcase the incredible work our entire team has done. We’re also looking ahead to what’s next on this journey.

Like true marketers, we have a video to kick things off—because why not?

(03:49)
[Video plays]

At TWE, we have more data than ever before, coming from various insights across the company. We needed a way to organize that data into a single access point for teams while also distilling it into actionable insights.

By leveraging AI, we quickly realized we could enable our teams to access insights more rapidly, effectively, and efficiently.

The most exciting thing about AI at TWE is that it provides a competitive advantage while allowing everyone to generate insights in their own way. This approach is helping us lead the industry.

One of the most beneficial aspects has been integrating AI with our usage and attitude study, Project Horizon. AI tools like RV and Viti use the same language we do when analyzing consumers and occasions, aligning with our market data. This ensures that when we identify opportunities, the insights are consistent with our internal planning and discussions with customers.

Currently, we’re launching a global brand, and this has been one of the most thoroughly tested concepts we’ve ever brought to market. At every stage, consumer insights have helped refine our proposition to ensure its relevance worldwide.

Many companies are still developing their AI policies, but Treasury Wine Estates is not only exploring AI but actively integrating it into daily operations.

They approached us while we were collaborating on other projects, asking if we could bring their AI agents to life—almost like real people. Using AI, we created two personas: Arvy and Viti. These represent the AI agents used to generate insights and drive innovation within the company.

A great example happened just last week. I was in the UK and got a question about sangria, which I didn’t know much about. Normally, I would have said, “Let me get back to you,” done research for a week, and then returned with an answer. But using RV and Viti, I quickly gathered insights and within 5–10 minutes, had key points on who drinks sangria, why, and what it could mean for our business.

This was a game-changer.

AI has helped simplify and make consumer insights more visible. Instead of 100-page decks, we now have insights that are easier to access and interpret. This makes information more accessible to our business and enhances how we share it with customers.

A key difference between TWE and other organizations is that TWE has embedded AI into its DNA. They’ve removed the fear of AI, enabling employees to embrace and understand the technology. With that cultural acceptance, there’s no limit to what they can achieve.

(07:53)
Adele: Reflecting on this project, it’s hard to believe it all started just eight months ago.

As part of the global consumer insights team, I quickly realized the overwhelming amount of data we had access to. In a central role, the first thing I wanted to do was understand what data we had, how to access it, and how to analyze different markets and consumer groups.

But there was no single place where all this data was consolidated. Everyone wants a one-stop shop for insights, but it didn’t exist.

We sourced about 18,000 documents scattered across desktops, SharePoint folders, and emails. The first step was organizing and centralizing all this data to make it accessible.

(10:06)
We developed Consumer HQ, a centralized repository for all our data. It allowed us to search for insights efficiently. For example, if we wanted information on the “Better for You” category, we could simply search "BFY" and access relevant documents instantly.

This eliminated the challenge of data overload and manual analysis. Instead of spending time compiling reports, we could focus on storytelling and brand building.

(11:42)
From there, we integrated AI to enhance our capabilities. The goal was to make data intuitive and actionable, leading to the creation of Viti and Arvy.

• Viti focuses on facts, ensuring data integrity, accuracy, and proper sourcing.
• Arvy is innovation-driven, generating creative ideas and helping us think outside the box.

These AI assistants act like team members. We wanted to humanize them so employees would feel comfortable using them. We debated calling them AI "agents" or "assistants" and ultimately chose assistants to emphasize their collaborative role.

(19:39)
Maddie: Since implementing this system, we’ve seen incredible results:

• Innovation: AI has helped generate fresh ideas, considering insights we hadn’t previously explored.
• Efficiency: Summarizing consumer feedback, compiling reports, and refining creative briefs now take seconds rather than hours.
• Consistency: Teams across different regions now speak the same consumer language.
• Self-service insights: Teams can now access insights without waiting on the insights team, freeing us up for strategic work.

This has been a game-changer.

(25:49)

Nicola: Fantastic! Thank you both. That was brilliant. It all looks so simple when you put it into a PowerPoint, but I know how much work has gone into this.

I’d love for you to take us back to the early stages—when you first had the idea. How did you get buy-in? Did leadership immediately support it, or was there a process to convince them?

(26:26)
Adele: Great question.

We’re fortunate to work in a curious, adaptable environment at TWE, which made a huge difference. We also partnered with GPT Strategic, our IT department, and Maddie’s team, ensuring everything was secure. That gave us the confidence to move forward.

We started small. Instead of asking for a massive budget upfront, we looked at what we would typically spend on innovation research and carved out a portion of that to experiment with AI.

Now, we have a test-and-learn budget every year, dedicated to trying out new ideas. That’s essential in marketing—having a space to experiment. We approached it as a small-scale initiative, got IT buy-in early, and collaborated behind the scenes before expanding.

Keeping it lighthearted and fun helped, too. We didn’t take ourselves too seriously in the early stages, which made the process more enjoyable.

(27:30)
Nicola: Fantastic. So, we talked about bringing the AI assistants to life—why was it important to give them personalities? What was that process like?

(28:12)
Maddie: It all came down to making them feel like extensions of the team.

Before they had names, it felt clunky—like we were talking about the AI insights tool, which sounded too robotic. Giving them names made them more relatable and approachable.

We also wanted the names to reflect TWE’s identity, so we chose ones related to wine. There are hidden meanings behind them—Viti comes from "viticulture," for example.

The goal was to make it natural in meetings to say, "Let’s ask Arvy," rather than, "Let’s consult the TWE Innovation AI Assistant." That felt too cold.

And of course, as marketers, we love branding—so we had to give them logos!

It’s been fun seeing them become part of everyday conversations. Even our CEO refers to Viti and Arvy as if they’re employees, which is pretty cool.

(29:18)
Nicola: That’s awesome! I was curious about the name Viti, so that answers my question.

A quick follow-up—who held the budget for this? Was it marketing or the data team?

(29:53)
Adele: It started as an innovation budget, but now it’s shared between marketing and insights.

That way, we all have skin in the game and a collaborative approach to experimenting, testing, and learning.

(30:28)
Rob: You had a pilot community—how long did it take from that test-and-learn stage to rolling it out company-wide?

(30:58)
Adele: We had three stages, though it wasn’t planned as seamlessly as it looks in hindsight!

1. Small core group (~5 people): The insights leads from different regions tested it first.
2. Expanded to the broader insights team (~20 people): Once we were comfortable, we brought in more team members.
3. Full business rollout: After proving its value, we rolled it out more broadly.

The entire process took about four months, which is quite fast. Once the insights team saw its potential, they were eager to expand access.

AI moves quickly, so we wanted to keep pace.

(32:07)
Nicola: That’s impressive! Now, in terms of training Arvy—since it pulls from broader sources—how do you control where it's pulling information from?

(32:46)
Maddie: Arvy is more open-ended, but Viti is knowledge-based and only pulls from trusted internal sources.

If Viti doesn’t have an answer, it won’t guess—it will simply say, "I don’t have data on that. Please upload a document if this information should be available."

We also made sure Viti cites references, so users can fact-check its responses. That’s critical, especially when using AI-generated insights for external presentations.

Arvy, on the other hand, is designed for creative, open-ended thinking, so it’s not constrained in the same way.

(33:54)
Heather: Has this reduced the need for primary research, or has it just made research more targeted?

(34:29)
Adele: More the latter—it’s made our research sharper and more efficient.

We’ll always need research, but AI helps us:

• Refine research briefs so we ask better questions.
• Improve creative briefs—our in-house agency is now producing some of the best work we’ve ever seen.

AI enhances our process rather than replacing traditional research.

(35:36)
SOA: Have you tracked any impact metrics—like time saved?

(36:10)
Adele: We didn’t set formal KPIs, but we had an internal goal: "Can we save 10 hours a week?"

That’s one full day per person, which frees up time for higher-value work.

We’re not tracking hours precisely, but we’ve noticed a major drop in ad-hoc requests, which tells us people are finding insights on their own instead of constantly asking the insights team.

We’re debating whether to formalize tracking or just keep focusing on improving efficiency.

(37:56)
Nicola: Any plans for a third or fourth AI persona? Maybe for sales or even a consumer-facing assistant?

(38:33)
Maddie: We just launched Penny, an AI assistant specifically for Penfolds (our luxury wine brand).

We’re also working on a consumer-facing AI assistant, which is really exciting. More to come!

(39:11)
Jimmy: How was the technology journey? Any major challenges?

(39:46)
Adele: Having IT’s support was crucial—security and trust are everything with AI.

Our partnership with GPT Strategic was also key. They helped us iterate and adapt rather than locking us into a rigid system.

The biggest challenge was organizing the data—centralizing it, structuring it properly, and making sure it was usable. That foundation made everything else possible.

(40:22)
Nicola: Speaking of organizing data—how did you collect 18,000 documents? Did you just send an email asking everyone to upload files?

(40:59)
Adele: That was one of the hardest parts!

We formed a small task force and made it fun—hosting a "hackathon" where we timed how many documents we could upload per minute.

The biggest challenge was aligning on categories and hierarchies—how to label and structure everything so AI could later retrieve it effectively.

We worked closely with IT to make sure it was stored in a logical, searchable format. That upfront effort made a huge difference.

(42:36)
Nicola: Amazing. Well, Adele and Maddie, thank you both so much!

Your journey has been incredible—not just in terms of technology but in transforming how TWE works. Who would have thought that when you started, even the CEO would be referring to Arvy and Viti as if they were employees?

Congratulations again!

To everyone here, we’ll be sharing the recording and an invite to our Slack community soon.