How Nexo Uses Agentforce to Resolve 62% of Crypto Support Cases Autonomously

Nexo operates a digital wealth platform serving a global base of cryptocurrency investors who expect instant, accurate answers to complex financial questions at any hour. Each month, the client care team handles more than 20,000 inquiries — many requiring real-time portfolio data and knowledge of regulations that vary by country. For a platform where trust and responsiveness are core to the product, slow or inconsistent support is a direct business risk.

Before deploying Agentforce, answering customer questions meant manually pulling information from multiple databases. Support agents had to cross-reference financial data, jurisdiction-specific rules, and internal knowledge bases before responding. The process was slow, error-prone, and impossible to scale overnight without adding headcount.

Nexo built and deployed Agentforce across its app and website using the Salesforce Agentforce 360 Platform. The team constructed the agent in two separate sandboxes — one global, one UK-specific — to validate behavior under different regulatory requirements before going live. They also built a custom Salesforce object that dynamically swaps terms like “withdraw” for approved regulatory equivalents, letting the agent interpret natural customer language without breaking compliance rules. Before launch, the team rewrote internal help articles to be readable by both humans and the AI model, giving Agentforce the foundational context needed to respond accurately.

The production deployment resolved 62% of customer cases autonomously and saved 2,600 hours of manual support work. When Agentforce encounters questions outside its permitted scope, it hands conversations to human agents without friction. Nexo’s Salesforce Product Manager, Nikolay Nedev, described the shift as moving from a rigid, over-restricted system to treating the agent like an intern: one that needs clear role definition and topic boundaries, but has genuine reasoning capability when properly guided.

Nexo plans to continue refining the agent as their product and regulatory environment evolve. The deployment demonstrates that high-stakes, compliance-sensitive industries can run effective AI support — as long as the data is clean, the content is written for both humans and machines, and the guardrails are precise rather than broad.