PhysiLink

What it does

PhysiLink is an Autonomous Educational Orchestrator designed to rescue students from passive learning. Unlike generic chatbots, it:

Psychological Profiling: It starts by assessing the student's cognitive style (Visual, Logical, or Kinesthetic) to tailor the entire learning experience.

Spatial-Temporal Video Reasoning: Using Gemini’s 1.5 Pro multimodal capabilities, it analyzes "Cause and Effect" within video lectures (e.g., explaining why an object moved at a specific timestamp).

The A-Z Narrative: It re-contextualizes lessons into a chronological story, starting from ancient philosophical roots to modern breakthroughs.

Automated Creative System: It generates precise, high-fidelity scientific diagrams and "Kitchen Lab" experiments that students can perform at home

How we built it

We leveraged the cutting-edge capabilities of the Gemini 1.5 Pro API:

The Brain: Used the 1M token context window to maintain a "Reasoning Trace," ensuring the AI remembers the student's psychological profile across hours of lecture content.

Vibe Engineering: Integrated Python execution environments to verify scientific simulations and ensure the accuracy of the physics being taught.

Frontend: Developed a sleek React & TypeScript interface featuring a dynamic "Historical Timeline" that syncs with video progression.

Creative Pipeline: Built a custom orchestration layer that directs Nano Banana Pro logic to create brand-consistent, labeled scientific assets

Challenges we ran into

Avoiding the "Wrapper" Trap: The biggest challenge was moving beyond a simple Q&A bot. We had to engineer complex system instructions to force the AI to act as a Creative Director and Spatial Analyst.Mathematical Rigor vs. Simplicity: Balancing the narrative story with mathematical accuracy was tough. We used LaTeX to represent equations like:$$\Delta x = v_0 t + \frac{1}{2} a t^2$$while ensuring the AI could explain the "story" behind the variables.Consistency: Keeping the "A-Z" structure intact across diverse topics required deep prompt engineering and iterative testing in Google AI Studio

Accomplishments that we're proud of

Successfully building a system that can analyze a 2-hour physics lecture and map it onto a 2,000-year historical timeline instantly.

Creating a Real-Time Teacher persona that adapts its tone—switching from "Persuasive Legal Logic" to "Simplified Visual Metaphors" based on the user's vibe.

Moving from "Simple Vision Labeling" to true Temporal Understanding, where the AI explains why something happened in a video, not just what happened.

What we learned

We learned that the power of Gemini 1.5 Pro lies in its Orchestration. It isn't just a language model; it's a reasoning engine that can bridge the gap between a visual frame, a historical fact, and a student's mental state. We also learned that "Vibe Engineering"—verifying code and logic in real-time—is essential for building trust in AI-led education.

What's next for PhysiLink

Our vision is to turn PhysiLink into a global AR/VR Classroom. We plan to:

Gemini Live Integration: Enable real-time voice-to-voice tutoring where students can interrupt "Newton" or "Einstein" to ask questions during a lecture.

Collaborative Study Rooms: Allow multiple students to explore the same A-Z timeline together in a shared virtual space.

Expanded Universe: Moving beyond Physics into Chemistry and Biology, creating a unified "Map of Human Knowledge."

Built With

• html
• script
• type