How Crocs built a retail superagent with Agentforce

Beyond a basic bot

With 129 million pairs of shoes sold in 2025 alone, Crocs has grown into one of the world's leading casual footwear brands. To support that growth, Crocs wanted to launch agentic shopping and service that could improve product discovery, answer sizing and fit questions, and reduce growing customer service queues.

Crocs quickly realized a basic FAQ bot – or a collection of disconnected AI agents – would not deliver the experience they envisioned. They wanted a single agent for customers that could guide product discovery, help shoppers find the right fit, resolve service questions, and know when to bring in a human – all from one continuous chat. 

To build it, they knew they’d need specialized expertise beyond their in-house resources.

Three teams, one shared vision

Crocs partnered with Salesforce’s forward-deployed engineers (FDEs) and Salesforce consulting partner Rosetree Solutions to bring AI superagent Rivet – powered by Agentforce – from concept to customer-ready in five months. "We saw Agentforce as a way to take advantage of conversational AI without having to build it ourselves," shared Feliz Papich, SVP, Digital Technology, Experience & Insights. "It let us get started quickly, learn what customers want, and avoid building all of that infrastructure on our own." 

Together, the three teams aligned on a phased implementation strategy, balancing fast time to value with a foundation that could support future expansion. That collaboration let Crocs move quickly without trying to solve every problem at once. Together, the team prioritized a focused launch centered on five high-impact use cases: product discovery, FAQs, order tracking, returns, and conversational checkout. 

“We designed the first release of Rivet to solve focused use cases so we could learn from real interactions before expanding,” said Craig Rosenbaum, Managing Partner, Rosetree Solutions. 

During the design and build process, Rosetree remained focused on the long-term vision, helping Crocs build an operating model that could grow beyond the initial deployment. Salesforce FDEs complemented that work with deep expertise in commerce architecture, testing, and implementation, allowing the team to deliver capabilities that would have been difficult for any one group to build alone.

Architecture built to grow

The implementation began with joint discovery workshops where Crocs, Rosetree, and Salesforce FDEs mapped common customer questions – including sizing and fit, product discovery, order tracking, and returns – directly to Agentforce capabilities.

“Crocs had an amazing vision for their first customer agent, and even bigger plans for the future,” Rosenbaum said. The team chose a superagent architecture, with specialized shopper and service agents coordinated by a single Orchestrator. 

“We decided to go the superagent route to ensure a foundation for scalability was put into place,” Rosenbaum explained. “Designating specialist roles from the beginning would prevent rework, and allow Crocs to deploy new use cases simply by adding a new subagent.”

Rosetree accelerated the design of the service agent by using existing single-order APIs to connect real-time order status, tracking, and return eligibility from IBM Sterling Order Management on day one. The architecture was intentionally designed for reuse, allowing future channels and agents to build on the same orchestrator, service agent, and trusted data foundation rather than starting from scratch.

Concept to customer in five months

“We were the experts on the service side,” shared Rosenbaum. “The FDE team led the Agentforce Commerce and shopper agent capabilities. Our skillsets complemented each other well, allowing us to go faster and deploy more robust capabilities than either team could alone.”

By sharing responsibilities across Crocs, Rosetree, and Salesforce, the team delivered a production-ready superagent in just five months while establishing an architecture designed to grow with the business instead of becoming another point solution.

That architectural decision is already paying off. Rosetree helped Crocs extend Rivet into Agentforce Voice in just seven weeks. “We were able to reuse the actions and instructions built for the Rivet service agent for Agentforce Voice,” said Rosenbaum.

From the outset, Rosetree also built in agent analytics, helping Crocs measure key metrics like autonomous resolution and FAQ accuracy so the team can monitor, refine, and improve Rivet over time.