Telephony control system with intelligent call routing

What is this patent about?

’253 is related to the field of computer-integrated telecommunications and intelligent switching architectures. Specifically, it addresses the technical challenge of efficiently routing real-time communications, such as telephone calls in a call center environment, by moving complex decision-making logic from high-level management systems directly into the low-level switching infrastructure. This shift aims to reduce latency and communication overhead while maintaining high-quality service levels.

The underlying idea behind ’253 is to replace rigid, rule-based routing with a dynamic combinatorial optimization engine that operates at the telephony server level. Instead of simply matching a caller to the first available agent or using basic least-cost routing, the system treats the assignment of multiple incoming communications to multiple potential targets as a global optimization problem. By evaluating the entire pool of available resources and pending requests simultaneously, the system can maximize total utility rather than just solving for the next call in isolation.

The claims of ’253 focus on a system and method that receive a plurality of communications and determine an optimum target for each through a non-trivial combinatorial optimization. This process utilizes data structures representing characteristics of at least three potential targets—such as agent skill sets—and matches them against multiparametric vectors derived from the communication sources. The independent claims specifically protect the mechanism of pairing sources and targets to achieve a maximum utility outcome based on defined optimization criteria.

In practice, the invention functions by gathering caller data through identifiers like ANI/DNIS or interactive menus to build a call characteristic vector. This vector is then compared against a table of agent skill profiles using a cost-benefit algorithm that can account for diverse factors, including agent compensation, predicted call duration, and even the long-term value of using a specific call as a training exercise for a trainee agent. This allows the system to adapt its routing strategy based on current system load, prioritizing immediate efficiency during peak times and personnel development during slower periods.

This approach differs from prior solutions by integrating the “intelligence” of the routing decision—such as Bayesian logic or neural network evaluations—directly into the switching architecture rather than externalizing it to a separate CRM or management platform. By performing these calculations locally on the telephony host, the system eliminates the non-deterministic latencies associated with external database queries. Furthermore, by moving beyond simple FIFO queues to a global pairing model, it avoids the “hot seat” problem and ensures that specialized agents are reserved for the tasks that most require their specific expertise.