ZeroSearch Scholar

A zero-search scholarship engine that tests whether less choice means better results.

Comment

Inspiration

Students often spend more time searching for scholarships than actually applying to them. The typical experience involves noisy listings, repetitive filters, and decision fatigue. During the hackathon, we questioned a core assumption of scholarship platforms: that users must search to discover opportunities. This led to a simple hypothesis: students might find better opportunities more efficiently if the entire search process is removed and replaced with automated, AI-driven recommendations.

What it does

ZeroSearch Scholar is an experimental scholarship discovery tool that removes traditional search functionality. Users do not browse or filter. Instead, the system automatically selects and ranks scholarships based on a profile vector, semantic similarity, and behavioral signals. The platform presents a single, curated feed of relevant scholarships along with an automatically generated timeline showing upcoming deadlines, progress stages, and recommended actions. The purpose is to test whether limited choice and algorithmic curation lead to faster and more confident decision-making.

How we built it

We began with a basic scholarship matching engine and converted it into a zero-search experimental platform. Key components include:

  • A React and TypeScript frontend built with Vite and styled using Tailwind CSS
  • Supabase for authentication, data storage, and user profiles
  • A PostgreSQL database for scholarship metadata and deadlines
  • An AI ranking layer that uses semantic similarity to score each opportunity
  • A minimal API layer built on Node.js and Express to serve matches and timelines
  • A deployment workflow using Netlify for the frontend and Supabase as the data layer

We designed the interface to minimize user actions and to emphasize automated recommendations.

Challenges we ran into

Building an experiment rather than a typical product introduced several constraints:

  • Removing search created initial confusion for test users who were accustomed to full control
  • Semantic ranking required careful tuning due to limited training data
  • Balancing simplicity with relevance was difficult since too few options felt restrictive, while too many resembled a traditional list
  • Deadline notifications affected users differently; some found them motivating while others found them stressful
  • Integrating multiple services under a tight timeline required compromises in architecture and testing

These challenges helped shape the final experiment and demonstrated how user expectations influence perception of AI-driven systems.

Accomplishments that we're proud of

We are proud that we were able to:

  • Turn a conventional scholarship finder into a functioning zero-search experiment within a short timeframe
  • Build an AI ranking flow that generated meaningful recommendations without requiring heavy user input
  • Create a timeline system that made deadlines more visible and actionable
  • Validate parts of the hypothesis using quick user testing
  • Explore a design philosophy that prioritizes reduction of choice rather than expansion of features

The project successfully demonstrates that scholarship discovery can be reimagined in unconventional ways.

What we learned

This experiment revealed several important insights:

  • Reducing choice can increase action, but only up to a point
  • Trust in algorithmic recommendations is fragile and must be earned through clarity and transparency
  • Deadlines and visual timelines have a measurable influence on user behavior
  • Rapid prototypes are effective for testing behavioral hypotheses and uncovering assumptions
  • Simplicity often requires more intentional design than complexity

We also learned that challenging common user experience patterns can generate valuable conversations about design and efficiency.

What's next for ZeroSearch Scholar

Future development will focus on improving the experimental model and exploring its practical potential. Planned next steps include:

  • Combining limited user control with AI curation so that users can influence results without using full search
  • Improving semantic ranking with more data points and additional context
  • Adding optional explanation layers to help users understand why certain scholarships were recommended
  • Introducing lightweight personalization methods that do not require detailed profiles
  • Expanding the experiment to test more hypotheses about choice, autonomy, and decision speed

The goal is to refine the idea further and evaluate whether zero-search discovery can evolve into a viable real-world approach.

Built With

Share this project:

Updates