Sales representatives have a challenging job. Especially when it comes to sales in IT, they need to learn about many aspects of products, pricing, market trends, customer needs, and many more topics. Finding the right information, and absorbing it as new knowledge can be very time consuming.

Large Language Models have the ability to digest enormous amounts of information and make it easily consumable through natural language. A great example application is ChatGPT, which made LLMs popular with the general audience.

This document describes the design for a web application that brings the benefits of Large Language Models to the sales representatives. The goal of the design is to help the sales person quickly evaluate example stories of successful customers, so that he/she can use that information when speaking to new potential prospects.

My role in this project was to create a high fidelity design. The intended functionality and interaction flow had already been decided before I joined the project. It was my task to make the design visually appealing, and to follow Oracle's look-and-feel standards.

The design consists of 2 screens. In the first screen, the sales rep can enter a natural language query to find existing customers that are relevant for a conversation with a new prospect that he/she is planning to have. For example: He/she may want to find examples of existing customers that were comparing Oracle and AWS. The language model produces summaries of customers that closely resemble the search term, and explains why each customer is a good match.

In the second screen, the sales representative can enter into a conversation about a particular customer story through a chat interface.

This screen contains a search field and search results. From this screen, the user can choose a particular customer story and talk about it in detail in screen 2 (chat).

I ensured following the company's look-and-feel by using the right UI elements, such as typography, icons, background elements and button styling.