Call Me Maybe

Call Me Maybe
Call Me Maybe Mockup

Inspiration

On average, people receive around 120 to 150 spam calls every month, with some even experiencing daily interruptions. These unsolicited calls can consume valuable time—between answering, listening to automated messages, or dealing with unwanted pitches, individuals often waste upwards of 10 minutes a day. This adds up to significant frustration and distraction, interrupting both work and personal life. Call Me Maybe offers a solution by redirecting unknown calls to an AI assistant, filtering out spam and ensuring you only engage with the calls that truly matter.

What it does

The AI assistant evaluates incoming calls by interacting with the caller, asking targeted follow-up questions to gather more context. It classifies each call as either 'Important' or 'Not Important' based on criteria such as emergency services, urgent medical appointments, and critical financial notifications versus non-urgent inquiries like telemarketing or surveys. If the call is deemed to be spam or unimportant, it is automatically declined. For calls classified as important, the assistant seamlessly redirects the caller to the appropriate responder.

How we built it

We designed the interaction between the AI assistant and the caller using a four-step process: caller audio, caller text, AI text, and AI audio. First, we record the caller’s voice and convert the audio to text using JavaScript Speech API. The AI then generates follow-up questions and responses using Generative AI based on the caller's input. This exchange continues until the AI can classify the call as either important, leading to a redirect, or not important, resulting in a decline. This efficient process facilitates effective call management based on the caller's needs.

Challenges we ran into

During development, our biggest challenge was accurately classifying calls from unknown numbers into two categories: important and not important. Achieving reliable differentiation often led to false positives and negatives, complicating the user experience. To solve this, we established criteria: important calls include emergency services, critical financial notifications, and urgent medical appointments, while unimportant calls encompass telemarketing, advertisements, and non-urgent inquiries. The AI asks follow-up questions after each prompt to gather more context without revealing the classification. Additionally, we faced challenges with user data security, gaining user trust, and integrating with various phone systems, all while adapting to evolving spam tactics.

Accomplishments that we're proud of

We are proud to have developed a potential app that effectively distinguishes between important and unimportant calls. This innovative concept has shown strong promise and garnered interest from peers and industry observers. Early feedback suggests that our approach could significantly enhance user experience by reducing interruptions from unimportant calls. We believe this solution addresses a real-world problem, helping users manage their calls more effectively. Overall, these accomplishments highlight our commitment to advancing call management solutions through this promising framework.

What we learned

Through this project, we learned the immense value of teamwork in creating and developing our idea. Collaborating effectively allowed us to brainstorm and refine our concept while building a clear product roadmap. Conducting thorough market research was crucial in understanding user needs and preferences, which informed our approach. Additionally, preparing and delivering our pitch helped us articulate our vision and gain valuable feedback. Along the way, we also gained important technical skills, which enhanced our capabilities and confidence in bringing our idea to life.

What's next for Call Me Maybe

Looking ahead, our next steps involve refining our app prototype based on feedback from initial presentations. We plan to conduct further market research to deepen our understanding of user needs and identify potential features that could enhance functionality. Additionally, we aim to develop a comprehensive testing plan to ensure the app's accuracy and reliability in sorting calls. We will explore opportunities for collaboration with industry experts to validate our concept and gain insights into best practices. Ultimately, our goal is to move toward a beta launch, where we can gather real user data and iterate on our solution to create a more effective call management tool.