-
-
Forge — Practice Strategy Builder Interface what about this
-
Forge — Intelligent Learning Platform Landing Page
Inspiration
We were inspired to explore the concept of Agentic AI and apply it in a real-world project. We wanted to build a system that doesn’t just respond to users but actively monitors progress and adapts over time based on performance logs. This is because we discovered that most apps or sites that train either punish users by breaking streak or just give a static (not evolving) strategy. Little or no sites/apps actually refines it's structure/strategy to meet user-specific learning methods. Our goal was to create a self-evolving practice engine powered by Gemini3 that continuously refines how users train, learn, and improve.
What it does
Forge is a Gemini3-powered autonomous strategy builder that creates personalized skill-building sessions, tracks performance over time, detects growth or stagnation, and automatically rewrites the practice plan when needed. It provides analytics dashboards, session tracking, email reminders and a visual timeline so users can clearly see their development journey.
How we built it
We built Forge as a full-stack web application combining React + Vite on the frontend and Python Flask on the backend. The React interface handles practice planning, analytics dashboards, timelines, and session logging, while Flask manages API routes, performance tracking, scheduling, and AI processing. We integrated the Google Gemini3 API to generate adaptive practice plans and analyze user progress. The system stores practice logs and uses scheduled tasks to evaluate performance over time, detect growth or stagnation, and automatically update practice system to keep training in a personalized and effective way.
Challenges we ran into
On the frontend, one of the biggest challenges was building the analytics section, especially implementing charts, making it responsive and making sure performance data was displayed clearly and accurately. Designing and structuring the timeline component manually was also challenging, as it required careful layout planning and dynamic rendering to properly reflect user progress over time. Balancing functionality with a clean, responsive UI while handling complex visual data was a key obstacle during development. While on the backend during development, the major issue faced was fine-tuning the Gemini3 AI, especially when trying to make the AI respond to different human wants. Because AIs (LLMs) often have a very vast language ecosystem, they tend to over emphasize and under-emphasize at times. Making Gemini3 respond according to a schema was the obstacle faced development. While during production, the main issue faced was configuring Render to run the containerized backend service.
Accomplishments that we're proud of
We’re proud that Forge works as a complete full-stack system — from the frontend interface with smooth animations, interactive charts, and timelines, to the backend handling data storage, performance tracking, Gemini3-powered adaptive AI, and most of all, the Agentic hands and legs_ one of which is directly connecting to user's email and Google calendar to set scheduled reminders for them. Seeing the entire system come together and function seamlessly was a major achievement. Finishing it in just a month also gave us firsthand experience of working under pressure and managing a full development cycle efficiently.
What we learned
From the Frontend and Backend, we learnt how to use Framer Motion more effectively to create smooth, meaningful animations that improve user experience. We gained a deeper understanding of Agentic AI concepts and how frontend interfaces can present adaptive systems clearly to users. We also improved our skills in data visualization, learning how to structure and display performance data effectively in charts and analytics dashboards. Most importantly, we learnt how to write cleaner, more maintainable React and Python code, organize components better, build structured & scalable UI and API systems and interact efficiently with LLMs to build Agentic Systems.
What's next for Gemini API App
We plan to expand Forge into a mobile app, allowing users to track their progress, complete practice sessions, receive real-time SMS and Whatsapp notifications/reminders and view analytics on the go for a more seamless and accessible learning experience.
Log in or sign up for Devpost to join the conversation.