Zefiro is an automated agent that motivates the e-commerce user to complete a purchase deal by negotiating price and conditions. Zefiro also acquires user information and preferences throughout the customer journey. It engages customers by adapting the purchase offer and recommending alternative products in order to match the user’s needs.
A weakness of web stores, compared to physical shops, is the lack of interactivity with online users. Zefiro aims reduce this gap by acting like a salesperson in a web shop. It acquires information to retrace the user’s profile and attempts to sell a particular item by engaging in a negotiation with the buyer. Zefiro elaborates the user profile by evaluating all interactions against a given ground knowledge of the web shop’s focus area. Hereby, Zefiro can quantify the richness of information acquired and recommend the most pertinent item according to customer’s needs.
Rather than a rule-based engine, Zefiro learns by itself how to maximize user conversions and sells. Zefiro is trained with reinforcement learning throughout thousands of simulations in order to gain expertise as seller.