Details
rundown is a GPT-backed solution for customizing and personalizing news based on user's proficiency levels on different topics and preferences. The product aims to utilize GPT to summarize complex news articles and present them in a digestible format and become a separate version of the existing Microsoft Start app. The product also has "clippy" which is a ChatGPT-backed chatbot which the user can access from top navigation menu to seek more information and explanation specific to the article or the topic. This project was made during the BigCo Studio program at Cornell Tech with Microsoft company advisors and stakeholders.
User Research
During our initial user research we tried to understand the needs and pain points of the users. We asked users how they consume their news, their sources, and their pain points. A lot of users said that they consumer their news from mainly social media platforms such as Twitter, Facebook, Instagram, etc. and some news websites. They also said that they sometimes don't understand certain complex articles and often loose interest. Users also highlighted their growing concerns regarding misinformation and biases in articles and their reluctancy regarding paying multiple subscription fees.
Process
Rundown operates with a straightforward process. It aggregates news from partner media channels and tailors summaries based on the user's familiarity with each topic. These personalized summaries are then presented as part of the rundown. Users have the option to delve deeper by reading the full articles. Additionally, they can choose to receive a general overview covering all topics and separate rundowns focused on specific subjects, each containing relevant news summaries.
Ideation & Solution
Based on our research, we decided to address the problem of long, lengthy, and complex articles. We also looked at Apple News as a news aggregator and set our vision to provide a comprehensive and elegant user experience with a competitive subscription-based and advertising-based revenue models. The product also aims to adopt a visual-heavy approach with a "story-like" design and infinite scroll in reverse chronological order for main pages.
If a user has a high proficiency level in finance, medium proficiency level in global politics, and low proficiency level in sports, the user will get complex summaries with technical jargon for finance news, medium difficulty level for global politics news, and easy-to-understand summaries for sports news.
Solution Validation
To evaluate our risks and the feasibility of our product, we decided to conduct three different experiments to analyze them. These experiments were tested on individuals based in the United States from 18 to 65 years old. A figma prototype was made to visualize and test the overall design of the product. Three experiments were designed to test our initial assumptions: digestible format for different proficiency levels, accuracy of summarized articles, and the spillover of the product. These experiments, research, prototype, and the overall product pitch can be access through the links.
Based on our experiments, users felt like they were able to understand the articles better by 20% and the users also felt that 98% articles were shortened with a 90% accuracy when we asked them to compare the real articles against the rundown summaries, and lastly, 75% of the users felt that they were willing to read more rundown articles.
