We are currently evaluating options for a US-based client who requires a customer-managed solution for email automation. The key requirements are as follows:

• A fully automated AI-based Email Bot solution capable of responding automatically to numerous emails from different shareholders and customers.
• The Bot should streamline email management by automatically processing incoming customer emails, identifying their intent, and responding with pre-configured templates.
• Elimination of manual review and the manual task of assigning emails to folders.
• Intent and topic detection with the ability to send pre-configured emails automatically.
• Capability to train the model for improved accuracy.
• Handling emails involving workflow processing and managing customer follow-ups.

We have proposed two solution options:

1. Pega Email Bot (Machine Learning and Rule-Based Learning)

   • Manual training required for the initial few months.
   • Concerns: In our past experience, the maximum confidence score achieved was 75% (0.75) after rigorous training and involvement of the Pega Decisioning team. The client's expectation is a confidence score of 85-90%. Could you please provide the highest confidence score that can be reached with the Pega Email Bot?

2. GenAI

   • Concerns:
     1. What is the highest confidence score that can be achieved with Pega GenAI?
     2. The client expects fully automated training. Can this be achieved without explicit API calls from Pega with the training data? Does this mean development effort is needed from the Pega side? Is our understanding correct?

We would appreciate your detailed response to these concerns to help us make an informed decision.