Viralytics

Opening Page
Input your interests and preferences
Receive videos similar to your preferences
Outputs ideas for your customized videos with hashtags and suggested video layout

Inspiration

Breaking into TikTok as a new creator is overwhelming. Trends move quickly, and while viral content is everywhere, understanding why something performs well and how to adapt it to your own style is difficult. We saw a gap between raw trend data and practical guidance. Viralytics was created to turn trend insights into personalized content strategies, allowing creators to stay authentic while aiming to be successful based on recent patterns in virality.

What it does

Viralytics is a trend-aware content assistant for new TikTok creators. By analyzing curated datasets of recent viral content, it helps bridge the gap between going viral and creating content that aligns with a creator’s style and goals.

Users can input preferences such as:

Primary niche
Interests
Preferred video style (Educational, storytelling, memes, etc.)
Whether they want to appear on camera
Effort level (low, medium, high production)
Desired video length

Using these inputs, the system generates multiple tailored video concepts inspired by current trends. Each suggestion includes:

A compelling hook
A structured script outline
Caption and hashtag suggestions
Recommended format and pacing
Suggested video length

In addition to generating ideas, Viralytics surfaces top-performing videos within the selected niche and highlights the style they use, such as educational, storytelling, aesthetic, or commentary-based formats. These videos are displayed directly in the interface and can be played within the platform, allowing users to study real examples alongside their personalized suggestions.

The goal is to give creators both inspiration and strategy. Instead of copying trends blindly, they can see what works, understand why it works, and adapt it in a way that feels authentic to their own content identity.

How we built it

We built Viralytics primarily using:

TypeScript for both frontend and backend logic
Node.js as our server environment
Prisma to manage and interact with our database
The Google Gemini API for generating structured, personalized video concepts
Limited integration with the TikTok API for exploring trend data

For this hackathon, we used a curated custom dataset of 2025 TikTok trends. Instead of focusing entirely on scraping large volumes of live data, we prioritized building a strong personalization pipeline that transforms structured trend data into actionable creative output.

Challenges we ran into

One of our biggest challenges was determining how to access reliable trend data. While TikTok provides visibility into trending content through its Creative Center, programmatically extracting that data is complicated due to permissions, authentication, and limited public API access.

We also faced:

Restrictions around scraping
Reduced visibility of certain trending metrics
Time constraints between building a scraper and refining our personalization system

Ultimately, we chose to work with a curated dataset so we could focus on implementing meaningful features rather than spending the majority of our time on data collection.

Accomplishments that we're proud of

We are proud that we:

Built a full personalization pipeline within a short timeframe
Successfully integrated generative AI to produce structured, usable content plans
Connected trend data with detailed user constraints such as no-face content, effort level, region, and video length
Designed an interface that shows real niche-specific viral examples alongside AI-generated suggestions

Rather than creating a generic idea generator, we built a system that translates trend intelligence into customized creative direction.

What we learned

Through this project, we learned that:

Accessing real-time social media trend data is more complex than expected
Structured data significantly improves the quality of generative AI outputs
Personalization matters more than raw trend exposure
Clear scope control is critical in hackathon environments

We also gained a deeper understanding of how viral patterns are often structural rather than random, and how those structures can be adapted without sacrificing authenticity.

What's next for Viralytics

Next, we plan to:

Implement automated daily or weekly scraping to continuously update our trend dataset
Expand from a curated dataset to larger-scale TikTok trend ingestion
Improve trend ranking and categorization models
Deepen integration with the TikTok API where possible
Expand beyond TikTok to include YouTube Shorts and cross-platform trend analysis
Enhance the chatbot refinement system for deeper creative iteration

Long term, we envision Viralytics as a real-time trend intelligence platform that empowers creators to grow strategically while staying true to their personal voice.