Zsombor Kukucska

Some UI improvements and bug fixes on the go!!! And also enhanced Dark/Light mode capability!!!

📌 Since Last Devlog…
Since our initial build and deployment, we’ve added essential groundwork for personalized learning through a Learning Style Quiz and integrated a simple user memory system via localStorage.
✅ New Features & Progress
🧩 Learning Style Quiz
• Built and integrated a 10-question learning style quiz
• Questions cover preferences like visual vs verbal, solo vs group, structured vs flexible learning
• On completion, the result is:
• Stored in localStorage (shown only once)
• Passed as learningProfile prop to the MainPage
• Displayed to the user as their profile
• Quiz is dynamically rendered with radio inputs and state-managed answers
🔁 State & Routing Improvements
• Quiz appears only on first app load
• React routing logic added: user is redirected to /quiz if no profile is found
• Profile presence is tracked via localStorage.getItem(learningProfile)
🧠 Profile Integration in MainPage
• MainPage.jsx now adapts based on learningProfile
(currently displayed, later will affect AI prompt behavior)
• Future AI prompt generation can use this profile to tailor tone, style, and depth
🔧 Technical Notes
• Cohere is still our AI backend (via /api/explain)
• App structure remains clean and modular:
• /components/LearningStyleQuiz.jsx
• /pages/MainPage.jsx updated with prop-based profile integration
• Routing handled via App-level state check (React Router planned for v2+)
🧪 Next Up
We’re focusing on features that can still be handled by the Cohere API:
• 🎯 Saving and retrieving full quiz results for future tailoring
• 📝 Basic study history log (session-based or saved explanations)
• 📆 Early calendar placeholder (non-synced)
🌟 Reflections
We now have the foundation to personalize learning experiences. Once we switch to a more advanced AI (likely OpenAI GPT-4o), we’ll unlock the ability to deeply adapt explanations, build schedules, analyze study patterns, and more.

📘 Devlog — July 20, 2025
✅ Milestone: First Public Deployment Complete!

Today marks a huge step forward for the MindPilot project — we successfully deployed the app publicly using Render! After solving a series of frontend and backend integration issues, the app now works flawlessly in production.

🚀 What was accomplished today:

🔧 Fixed all production build issues (e.g. vite build failures, StaticFiles directory, missing modules)
⚙️ Finalized build.sh script to automate frontend & backend integration
🧠 Verified the full explain flow works end-to-end with Cohere's API
✅ Functional features working in public:
Topic input
Subject selection dropdown
Detail level selection (simple / normal / detailed)
AI explanation generation
Save explanation to localStorage
Download explanation as .txt
🧪 Testing Outcome:

The live site correctly loads the React app.
Explanations are generated and displayed with Markdown rendering.
Saving and downloading works seamlessly.
No frontend or backend errors observed post-deployment.
📦 Deployment stack:

Frontend: React + Vite
Backend: FastAPI + Cohere API
Deployment: Render (build & runtime), build script included
Static hosting: Vite build output served via FastAPI with StaticFiles
🗒️ Next Steps:

Launch public testing (gather user feedback)
Begin implementing MindPilot v2 roadmap:
🔍 Search result highlighting
🧠 Learning profile features
📅 Smart calendar and study planner
📈 Visual progress tracking
💡 Topic relationship visualizer

🛠️ Devlog – June 24, 2025
📌 Summary:
Today, we focused on making previously saved explanations easier to navigate by implementing a search feature on the History page of MindPilot. This significantly improves usability, especially as the list of saved explanations grows.
✅ What Got Done:
🔍 Search Functionality for History Page:
Users can now search their saved explanations by topic keyword.
The search is real-time and filters the list instantly based on input.
Ensures better access to relevant explanations without having to scroll through the entire list.
🧠 Planning and Vision Refinement:
Reconfirmed the direction of MindPilot as a smart learning assistant.
Prioritized features for the upcoming 2.0 roadmap (e.g., smart calendar, learning style profiling).
Discussed positioning the app as a year-end exam project, portfolio piece, and possible thesis foundation.
🧩 What’s Next:
Start implementing filters by subject and sorting by date on the History page.
Begin groundwork for the Study Planner module powered by AI and integrated with a user-managed calendar.

🛠️ Devlog – June 23, 2025
Today's focus: UX improvements, dark mode support, and a new History page
Today’s session was all about polishing the user experience and adding more structure to the app. Here's a summary of the progress I made:
✅ Key Improvements
🌗 Automatic dark mode support
MindPilot now adapts to the user’s system theme. If your OS/browser is in dark mode (like Safari on macOS), the interface adjusts accordingly — ensuring proper contrast, readable text, and overall visual consistency.
📂 New History Page
I created a dedicated History page that shows all previously saved explanations. This marks a big step toward treating MindPilot as more than just a one-time assistant — it’s becoming a true personal knowledge companion.
🛠️ Component refactoring
I started organizing the app structure by creating a pages/ directory and splitting components like the main interface and the history view into separate route-driven pages using React Router. It’s cleaner, easier to manage, and future-proof.
🎨 Minor UI polish
I fine-tuned contrast and layout spacing so that elements like the prompt, dropdowns, and output text don’t clash in either dark or light modes.
💡 Reflections
While the changes today were mostly visual and structural, they add up to something important: comfort. Users will now feel more at home using the app — whether it’s at night on a dark screen or when reviewing past learning moments.
🔜 Next Steps
Add advanced filtering and search in the History page
Optionally allow editing or deleting saved explanations
Add animations or loading transitions
Begin working on subject-based progress tracking

Devlog – 2025.06.21

Today I worked on enhancing the MindPilot app by adding a detail level selection feature for AI explanations. This allows users to choose how simple or detailed they want their explanations to be, making the app more flexible and user-friendly for different learning needs.

The detail level selector was successfully implemented as a dropdown, and I connected it to the backend so the AI can tailor responses accordingly. This involved updating the frontend state management and adjusting the API calls to include the chosen detail level.

Additionally, I considered the idea of creating a separate page for this new functionality to keep the UI clean and organized. However, I decided to postpone this to tomorrow to maintain focus and avoid overcomplicating the current workflow.

Overall, today's work improved the app's usability and set the stage for further AI-powered customization options. Tomorrow, I plan to revisit the multi-page structure and start implementing additional AI features.

📅 Devlog — June 20, 2025
Today we added two key features that make MindPilot even more useful and flexible for learning.
✅ What’s New
📚 Multiple Subject Support
Users can now select the subject area (e.g., Physics, History, Programming, etc.) for their explanation. This allows the AI to generate more relevant and accurate responses tailored to the context of the topic.
💾 Save Explanation Feature
Added a “Save Explanation” button that allows users to download the generated answer as a .txt file. This makes it easy to save useful explanations for later review or offline study.
🚀 Why It Matters
These two features take MindPilot a step closer to becoming a versatile AI learning assistant. The subject selector improves the quality of explanations, while the ability to save results helps students stay organized and build their own personal knowledge base.

🛠️ Devlog – June 19, 2025
Today’s progress on MindPilot
✅ Backend successfully integrated with Cohere API:
We switched from OpenAI to Cohere due to API quota issues and integrated it smoothly into the FastAPI backend. The .env setup now safely handles API keys.
✅ Frontend-Backend Communication Established:
Our React frontend now successfully sends requests to the FastAPI backend and receives real-time topic explanations.
✅ Full UX cycle working end-to-end:
Users can now enter any topic, click “Explain,” and get a detailed explanation powered by AI, displayed directly in the interface.
✅ Stylish UI Enhancements:
Replaced text emojis with React Icons for a cleaner, more modern design. Added dynamic loading feedback and polished input/output layout to improve clarity and engagement.
🧠 What I learned:
How to debug CORS and API key issues in FastAPI
Connecting React frontend with a real-time backend
Using environment variables securely
UI/UX improvements with React Icons
🔜 Next steps:
Add explanation history
Personal session dashboard
Prepare mobile-responsiveness
Continue devlogs and start documenting feature roadmap

Dev Log – MindPilot – 2025.06.18
Today, significant progress was made on the backend of the MindPilot project. I successfully set up a FastAPI server that handles API requests for AI-powered topic explanations. I integrated the Cohere API using their Command R+ model to generate clear, simple explanations suitable for learners.
Key tasks completed:
Configured environment variables securely with a .env file for API key management.
Built the /api/explain POST endpoint that receives a topic and returns an AI-generated explanation.
Ensured proper CORS middleware settings to allow smooth communication with the React frontend running on localhost.
Tested the backend thoroughly using curl, Postman, and FastAPI’s interactive docs, confirming error-free responses.
Resolved common errors related to API key handling and Cohere client usage, improving the stability of the backend.
Next steps:
Connect the backend with the React frontend to create a seamless user experience.
Expand the AI capabilities by adding new features such as personalized progress tracking.
Begin planning for additional languages and multi-subject support.
Overall, today laid a solid foundation for the core AI functionality of MindPilot and set up a stable backend infrastructure for future development.