The legal matters handled by social secretariats, such as the law on educational leave or issues related to payslips, are very diverse and often complex, and the accuracy of automatically generated responses is imperative.

To address this, Sagacify has developed a system based on multiple AI agents, each connected to a specific legal area, procedure, or circular and designed to master that area. When a question arises, the main AI agent identifies the subject(s) involved and delegates the processing to the relevant specialized AI sub-agents, which are then responsible for providing the most accurate and correct answer. When this is not sufficient and the question requires a response following a specific processing method, it is possible to define specific “chains of thought” that follow a strict business logic to ensure accuracy, even in the case of complex questions.

Thus, the Mailbot system is flexible, in the sense that it can handle an increasing number of topics, and scalable, in that the knowledge connected to the specific AI agents can be gradually expanded, and new, stricter “chains of thought” can be developed when necessary.

Chronologically, it is advisable to identify the subjects in incoming emails that represent the largest volume and to focus the development of the Mailbot system on these areas, then gradually evolve the system towards topics with lower volumes. If full automation isn’t possible, it compiles the relevant details along with clear next steps to guide the team’s response, with all outputs reviewed by UCM staff before being sent. When a request is too complex or unsupported, the system automatically forwards the email to the relevant staff member for follow-up.

This system therefore demonstrates great flexibility and adaptability, and can easily keep up with the evolution of topics addressed in client queries. UCM can thus reduce its backlog of unanswered questions and focus its efforts on providing faster and higher-quality customer service.