How Marketers Are Putting GenAI to Work With Skai’s Celeste at ShopAble 2025

ShopAble 2025 is an annual commerce media event hosted by Skai, gathering marketing leaders, industry innovators, and brand executives to explore emerging trends and practical strategies driving the future of commerce. Held in New York, ShopAble offered candid insights, actionable takeaways, and real-world stories from the forefront of retail media, generative AI, and omnichannel marketing.

One of the standout sessions this year was Your Future Today: How Marketers Are Putting GenAI to Work with Celeste, led by Skai’s Guy Cohen, Chief Product Officer. Cohen shared how Skai’s Celeste, the first GenAI marketing agent purpose-built for commerce media, was developed, what marketers are already accomplishing with it, and how it will help evolve marketing teams in the months ahead.

He was joined on stage by Kelly Gerrard, Director of E-Commerce at Marshall Associates, one of Skai’s early partners for Celeste, who shared firsthand how her team is using Celeste to drive new levels of efficiency, insight, and strategic impact.

You can watch the full session recording here.

How Celeste came to life — three requirements

Cohen: “In early 2023, we created a concept of a marketing agent—a virtual team member that should take repeatable tasks from marketers and allow them to do what only humans can do: strategize, innovate, ideate, and dream. We shared this concept with a few clients, and their response was really positive.

“But our clients also had a few requirements. First, this technology should work in a closed and secure environment—no data can be shared. Second, it needs to be trustworthy—no Disney-in-London type hallucinations. And third, clients weren’t interested in working with a generic agent. They wanted something tailored to our industry, knowing all the complexity, nuances, and years of knowledge that have been gathered.”

How marketers are already using Celeste

Cohen: “We found that the questions clients ask naturally fall into three main categories.

“The first category is summarization—from quick snapshot questions like, ‘Summarize last week’s performance and identify underperforming campaigns,’ to executive briefing questions like, ‘Create a QBR presentation summarizing last quarter.’

“The second category is around insight. Clients are asking Celeste to learn from their history—questions like, ‘Analyze last year’s Prime Day performance and suggest a better plan,’ or, ‘Is there a correlation between my Facebook brand awareness investment and my Amazon sales?’

“The third, and probably the most powerful category, is about making strategic decisions—questions like, ‘I have an additional 10% of budget. What is the optimal way to invest it to get the best outcome?’”

What Marshall Associates is seeing with Celeste

Gerrard: “At first, we started with basic questions—asking things like, ‘Can you summarize this dataset?’ Something that myself or my team could have done, but testing the efficiency and accuracy.

“In the last few weeks, we’ve been asking more complicated questions. We’ve been trying to trick her a little bit to see if she can pick up what we know or data trends that we’re aware of? Or can she bring something we haven’t thought of yet? That’s where it starts to feel like using her as an associate team member, rather than just a data summary.”

Where Celeste is driving the most value

Gerrard: “A big part of how we quantify success is internal efficiencies—not just optimizing campaigns for clients, but making sure our teams are working smarter. We manage on average 350,000 SKUs and 1 million keywords. That’s a lot of data to digest and catch trends.

“Instead of taking myself or someone on the team three hours to identify a data trend, it takes 10 minutes. One example: in a water sports category, we asked Celeste why CPCs were down and the conversion rate was up. Celeste surfaced that search terms including ‘youth’ or ‘kids’ were outperforming adult terms, months ahead of the core season. That gave us a competitive advantage we could leverage.”

How Celeste will reshape marketing teams

Gerrard: “It lets my team do what we do best—be an extension of our brands. It lets us spend less time on thousands of rows of data. It’s that superpower to let us focus on client work, building relationships, creating strategy—using Celeste almost as our back-pocket data summary and assistant.”

What’s next for Celeste

Cohen: “Everything you saw today is just version one. Our vision is to create an agent that is so deeply embedded in your business that it feels like a natural extension of your team—an agent that doesn’t just answer questions, but understands your goals, works alongside you, learns from your data, adapts to your workflow, and brings your organization’s knowledge to life.

“You’ll deploy multiple agents—each one with a specific role, always on, always learning, always focused on impact. Imagine an agent who reviews brand performance every Monday morning and raises flags. One that prepares a Prime Day strategy weeks in advance, or that generates a QBR presentation after the quarter ends, beautifully formatted and packed with insights.

“This isn’t a distant future. This is the next phase of Celeste. We’re not just building tools—we’re building an intelligent teammate.”

Key takeaways for marketers

AI marketing agents are already here. The opportunity is no longer theoretical—early adopters are already putting agents like Celeste to work for strategic impact and efficiency gains.

Three key use cases are emerging. Summarization, insight generation, and strategic decision-making are the most common—and most valuable—ways marketers are using GenAI agents today.

Internal efficiency is a measurable win. Celeste is already saving marketing teams significant time on data analysis and reporting—freeing up talent to focus on higher-value work.

AI agents will accelerate marketing teams, not replace them. The goal is to embed agents as team members, allowing human marketers to do more of what they do best, supported by AI-driven insights and execution.

What’s next will happen fast. Agent capabilities will evolve quickly, with use cases expanding from QBR preparation to proactive campaign management, optimization, and planning across marketing cycles.

The shift from tools to teammates is already underway, and the next phase of GenAI will center on how deeply marketing organizations can integrate these agents into daily workflows.

You can watch the full session recording here.