jili tutorial

Resolution Revolution: Why Customer Service Must Move Beyond Containment

By jili tutorial · September 27, 2026

Resolution Revolution: Why Customer Service Must Move Beyond Containment

Introduction: The Promise vs. Reality of AI in Customer Service

The artificial intelligence revolution promised a new era for customer service—one where customers no longer dread long hold times, repetitive menus, or unhelpful chatbots. Instead, they would interact with always-available AI agents capable of resolving any issue instantly. Yet for many organizations, especially in regulated industries like iGaming, the reality has fallen short. Automation tools are often designed to contain inquiries rather than resolve them. A customer asking about a delayed withdrawal may be directed to an FAQ page, left frustrated, and eventually forced to abandon the issue altogether. The operator may even count this as a successful interaction—despite the unresolved problem—creating a ticking time bomb of repeat contacts, customer churn, and, in gambling, missed signs of financial distress or responsible gambling concerns. This containment-first approach poses serious risks to player protection and regulatory compliance.

The Cost of Containment: Why Resolution Matters

When customer service automation merely deflects rather than solves, the consequences ripple across the entire business. Consider these hidden costs:

True resolution—where the customer’s problem is fully addressed in one interaction—is the only sustainable goal. Yet many generic AI tools are not built for the depth required in specialized industries.

Why Generic AI Falls Short in iGaming

Raphie, a New Jersey–headquartered AI platform originally founded as Conduet in 2019, has built its entire mission around this gap. Its co-founder and CEO, Justin Heath, explains: “A big premise on which we have built Raphie is to completely avoid that situation where it feels as though you are talking to a robot that does not know what you are asking.”

The core issue is that approximately 90% of incoming customer contacts in iGaming require account-specific information. A player asking about a withdrawal’s status may need to know whether a fraud review is holding the process, what documents are required, or an accurate payment timeline—not a general policy link. A generic chatbot that lacks integration with the operator’s wagering system will fail.

Similarly, queries about bet settlements, bonuses, or game rounds demand deep industry knowledge. For example, a customer questioning why a parlay lost needs the system to break down each leg and explain the specific result—not just recite terms. Without access to wagering data and account context, generic AI is essentially blind.

The “Context Layer” Solution

Raphie addresses this through what it calls a context layer—a system that integrates directly with the operator’s player account management (PAM), wagering, and customer service platforms. Before responding, Raphie retrieves relevant account and transaction data, then applies the operator’s specific procedures. This ensures every answer is personalized and actionable.

Heath notes that operators often initially dismiss the need for a specialist platform because they already use tools like Salesforce or Zendesk with embedded AI. “They expect the 70% or 80% automation advertised on the box, but then find it is more like 5% or 10% because that is the volume you can respond to without the wagering-technology integration.” The difference is stark: generic AI advertises high automation but delivers little more than FAQ deflection.

Raphie’s Approach: Accountable Automation

Raphie’s philosophy, which it calls “accountable automation,” redefines success. It is not about maximizing the number of conversations handled without a human agent, but about ensuring every decision is measurable, auditable, and made within an operator-approved scope. This approach has produced tangible results: one US sportsbook client reported customer satisfaction scores above 90% among players served by fully automated service.

Three Modes of Deployment

Raphie offers three deployment modes that can be used independently or blended according to an operator’s needs:

  1. Full Automation – The Raphie AI Agent handles supported interactions end-to-end. When human intervention is required (e.g., for complex or sensitive issues), the system hands off transparently.
  2. Co-Pilot – The AI suggests responses to human agents, who can use them as written or edit them. This accelerates agent workflow while keeping a human in the loop.
  3. Specialist Human Agents – As a standalone service, Raphie provides trained human agents based in Jersey City and the Philippines to cover customer service, risk, and payments support.

This flexible architecture allows operators to start conservatively and expand automation as trust builds.

Measuring True Resolution: Beyond Vanity Metrics

Traditional automation metrics often count a conversation as “successful” simply because it ended without a human handover—even if the customer abandoned the chat in frustration. Raphie rejects that approach. Its resolution calculation considers:

Crucially, Raphie uses large-scale quality assurance to detect conversations that were abandoned before reaching a proper conclusion. Heath explains: “An abandoned conversation is a very negative outcome. It means the customer has been going around in circles or has not received the response they need. We assess where a customer has dropped off before the conversation has fully run its course rather than treating that as resolved.”

This rigor ensures that every interaction counted as “resolved” truly is. For co-founder and COO Alex Jones, an even more telling sign of quality is when customers don’t realize they’re talking to a bot. “We see people wishing Raphie a good night and signing off because they have enjoyed such a natural dialogue,” he says.

Case Study: BetSaracen’s Success with Raphie

Arkansas sportsbook BetSaracen illustrates how Raphie’s modes can work together in practice. The operator began using Co-Pilot over a year ago and later introduced Full Automation. In its first eight weeks, Full Automation achieved:

These results were not achieved overnight. BetSaracen followed a phased approach to ensure quality at every stage.

A Controlled Rollout Strategy for Operators

Raphie recommends a four-stage process for introducing automation, designed to minimize risk and maximize trust:

  1. Assessment – Analyze the operator’s typical inquiries, workflows, and existing technology stack.
  2. Shadow mode – Raphie generates responses but they are not sent to players. This allows testing of accuracy and relevance.
  3. Category-by-category rollout – Automation is enabled for specific inquiry types (e.g., withdrawal status, bet settlement) one at a time.
  4. Continuous optimization – As products, promotions, and player behavior change, Raphie adapts iteratively.

“We usually start with a narrow automation scope and then expand it as we prove the resolution and the human-like response,” says Heath. This gradual approach also allows operators to add new actions to Raphie’s capability. For example, a bet-settlement inquiry might reveal that an event has finished but the market remains unsettled—Raphie can then alert the trading team while updating the player. It can also notify a game provider about a stuck round or unlock an account when the operator’s criteria are met.

When Human Intervention Is Required

Raphie recognizes that resolution sometimes requires human judgment. Enquiries are handed over when they:

Heath gives a concrete example: “If somebody asks about a pending withdrawal and then says they need the money to pay their rent, that type of flag is escalated straight through to the relevant team.” This balance between automation and oversight is central to accountable automation.

The Future: Accountable Automation in Europe’s Regulatory Patchwork

Having proven its model in North America, Raphie is now expanding into European markets. This move is driven by strong evidence that its resolution-led approach reduces costs without compromising service quality—a particularly attractive proposition given rising gambling taxes and operational pressures across the continent.

However, Europe is not a single market. Raphie describes it as a “regulatory patchwork” where underlying knowledge of payments, bonuses, and KYC remains relevant, but the compliance and escalation layer must be tailored jurisdiction by jurisdiction. Player protection rules, responsible gambling triggers, disclosure requirements, and audit expectations all differ between countries.

European operators already run sophisticated in-house customer service operations. They understand the challenges intimately. But as cost pressures mount, the appetite for AI that truly resolves—rather than contains—is strong. “European markets are under far more cost pressure, so the desire to move quickly with AI seems to be there,” Heath notes.

Conclusion: Moving Beyond Containment

The AI revolution in customer service is not a failure of technology—it is a failure of implementation. Too many organizations settle for automation that deflects, defers, and deflects again, mistaking containment for resolution. Raphie’s success demonstrates that a specialist, context-aware, and accountable approach can deliver both high customer satisfaction and significant cost savings.

As the industry evolves, the winners will be those who prioritize genuine resolution over vanity metrics. For operators in iGaming and beyond, the message is clear: stop counting abandoned conversations as successes. Start building automation that your customers won’t even realize is automated—because it solves their problem on the first try.