An interface for visualizing a chatbot's knowledge.


persona definition -> requirements gathering -> card sorting -> paper prototyping -> needs validation -> high-fidelity prototyping -> usability testing -> beta ship -> ...


Matt Tharp, Jon Linton, Alan deLespinasse, Ira Ladson


2019 / 3 months

Problem Statement

Gamalon's natural language engine can ingest the contents of a website and use this information in a conversational context. The information lives in a backend database where, in its raw state, it is not human-readable. Therefore, there is a need to create an interface for Gamalon's clients, who are usually marketing professionals without technical data science backgrounds, to access the data and make changes to it.


IdeaBase visualizes a database of natural-language data in a human-readable format. It organizes titbits of information into a hierarchy that is familiar for marketing professionals, and it uses the principles of progressive disclosure to allow marketers to go in and make changes as they see fit.

Selected Assets

A selection of assets from the prototyping process are presented here.

[Figure 1] An early wireframe showing a high-level view of the chatbot's knowledge base.

[Figure 2] A zoomed-in view showing the details of a particular nugget of knowledge.

[Figure 3] A zoomed-in view showing data about how often a particular piece of information is invoked during chat.

[Figure 4] An early wireframe showing how the user might add to the chatbot's knowledge by answering questions that the bot cannot currently understand.

[Figure 5] An animation showing a high-fidelity prototype of the IdeaBase interface.