Qlearo

Qlearo generating explainer math videos using Manim
Social mode used to study about Cluely's implications
Extreme mode used to summarize Apple's performance

Inspiration

As an computer science student and builder, I often found myself in the sea of browser tabs - LLMs for answer, Google for sources and PDFs and YouTube for explanations. Despite having multiple AI tools in the market, they were not aware of the latest news and I realized I was still lacking clarity.

I wasn't alone - millions of students and researchers face the same problem daily: AI gives answers, but not understanding.

That's when the idea of Qlearo was born : an AI tool that searches for sources across the internet in real-time, summarizes them with in-line citations, offers related questions that the users might ask and also generate visuals to understand a topic.

What it does

Qlearo is an anonymous AI search & learning engine that searches and teaches you with animated visuals, citations, and interactive breakdowns.

Here’s what Qlearo does:

Search & Research: Ask any question and get structured, cited, and context-aware responses choosing between dozen LLMs you want.
Deep Research (Extreme) Mode: Explore complex topics with depth, follow-up questions, and academic-level insights.
Animated Video Explanations: Generate visual explanations using the Manim library, like Khan Academy to understand tough concepts easily.
Academic & Financial Data Access: Pull real-time information from sources like arXiv, Semantic Scholar, Yahoo Finance, Alpha Vantage, and more.
Voice Interaction: Use real-time conversational agent powered by Elevenlabs for hands-free learning.

How I built it

I built Qlearo using a modern AI tech stack focused on performance, clarity, and interactivity.

Frontend: Developed with Next.js and Tailwind CSS for a fast, responsive UI.
UI Components: Leveraged shadcn/ui and Lucide icons to create a clean, minimal interface.
LLM Integration: Used multiple LLMs you can choose from including Open AI, Anthropic, Gemini and other open source models (it currently supports only Gemini, Fireworks & Groq models) with custom prompts to deliver structured, multimodal answers.
Search APIs: Connected to Tavily, Serper and Jina for searching through the internet.
Financial APIs: Used Alpha Vantage and Yahoo Finance for real-time finance data.
Voice Interaction: Implemented real-time conversational agent responses with ElevenLabs.
Video Rendering: Used the Manim python library to programmatically generate animated video explanations for complex concepts.

Challenges I ran into

Building Qlearo in 30 days came with plenty of challenges:

Combining multiple APIs: Coordinating between different APIs (LLMs, search engines, academic databases, financial data providers, etc.) was tricky. Each had different response formats, rate limits, and quirks that needed normalization.
Video generation with Manim: Generating animated explanations programmatically using Manim required fine-tuning scripting, timing, and ensuring videos rendered quickly enough to be usable on demand.
Custom UI complexity: Designing a clean, distraction-free interface that could handle complex output like graphs, citations, videos, and search threads without overwhelming the user, was a real UX puzzle.
Prompt engineering: Ensuring consistent, structured, and context-aware responses from the LLM required significant experimentation with system prompts and formatting techniques.

Accomplishments that I'm proud of

I built a full-stack AI search and learning tool complete with real-time search, structured AI answers, academic/financial data retrieval, animated video explanations, and voice interaction.
I successfully integrated Manim to generate dynamic, educational video explainers programmatically something I’d never done before.
I designed and implemented a prompt strategy that consistently returns structured, multimodal answers (including sources, graphs, and charts) with minimal hallucination.
I pulled together complex UI components and multiple APIs into a seamless, intuitive user experience without relying on templates or no-code platforms.

Qlearo started as a solution to my own frustration with fragmented AI tools, and I’m proud to have brought it to life in such a short time with full functionality and polish.

What I learned

I deepened my understanding of prompt engineering, especially how to guide LLMs toward structured, multimodal, and accurate outputs using advanced system prompts.
I learned how to integrate and normalize data from multiple sources — including AI search engines, academic APIs, and financial APIs — while maintaining performance and reliability.
I explored the Manim animation library for the first time and discovered how to convert abstract concepts into animated video explanations dynamically.
I learned how to design better user experiences for complex AI outputs — especially how to display citations, graphs, charts, and longform answers in an elegant, readable interface.
Above all, I learned how much can be accomplished under pressure when focused on solving a real problem with clarity and creativity.

Building Qlearo was both a technical challenge and a creative journey — and it’s made me a much stronger AI developer.

What's next for Qlearo

Qlearo is just getting started. I’ve validated the core concept that will take it to the next level.

Here’s what’s next:

Smarter Knowledge Graphs: Visual maps that connect concepts, questions, and sources to help users see how ideas relate across domains.
Better Memory + Context: Add persistent session memory, so users can build long-term research threads and Qlearo can grow with them.
More Data Integrations: Expand academic and financial sources to include JSTOR, Scopus, SSRN, TradingView, and SEC filings.
Collaborative Workspaces: Enable real-time collaboration with teammates or classmates inside research "Spaces" where users can "discuss" with Qlearo on research topics.

Qlearo’s mission is to turn passive information into active understanding and this is just the beginning.