Overview

I am building IntelliPath AI, a web platform designed to help users make consistent progress across any area of life—career, finance, technical skills, or personal growth. The app combines AI-powered chat, YouTube transcript analysis, and structured note generation to provide clear, actionable takeaways from video learning content. Users can save content, search past work, and organize videos/notes into categorized folders. Why this matters: The internet has an abundance of self-improvement and technical learning resources, but it’s overwhelming and time-consuming to filter through. IntelliPath AI creates a personalized productivity system that converts passive watching into active knowledge retention and structured learning.

Key Features

• Transcript Analysis & Summarization
  • Integrated AI models to parse YouTube transcripts (with fallback to external STT if disabled).
  • Automatically generate structured notes: key points, practical applications, and action steps.
  • Notes adapt to user-defined goals (e.g., “I want to master the 48 Laws of Power”).
• AI Assistant Chatbox
  • Users can ask questions about video content or their notes.
  • AI provides context-aware answers, explanations, and examples.
  • Chat history is persistent and searchable.
• Sidebar Navigation (ChatGPT-style)
  • Recent conversations automatically titled by keyword + prompt.
  • Options to rename, share, or delete conversations.
  • Saved folders auto-organize by subject/goal; users can also create custom folders.
  • Popup window for folder contents with search and scrolling support.
• Monetization & Plans
  • Free plan with limited history/summaries.
  • Premium plan (Stripe integration) with 14-day trial, unlimited summaries, advanced notes, and folder management.
• Cloud Storage Integration
  • Supabase database and storage ensure persistence of conversations, notes, and saved videos.
  • OAuth 2.0 (Google, LinkedIn, Apple) for account authentication.
• Automated Categorization
  • Saved content is auto-tagged by theme and learning goal.
  • Enhances search and helps build thematic knowledge libraries.

Technologies Used

• Frontend: Next.js (React), Tailwind CSS, shadcn/ui components, lucide-react icons.
• Backend: FastAPI (Python), Node.js API routes for handling video/transcripts.
• AI Models: OpenAI API (chat + summarization), Whisper/AssemblyAI for transcript fallback.
• Database & Auth: Supabase (Postgres, storage, OAuth 2.0).
• Payments: Stripe (plans, billing portal, free trial).
• Cloud: Vercel/Tempo for hosting and CI/CD integration.
• Integrations: YouTube Data API for metadata, transcript parsing libraries.

Challenges and Learnings

• Transcript reliability: YouTube’s transcript API often failed when captions were disabled. To address this, I implemented a fallback transcription system (Whisper/AssemblyAI) to ensure coverage for any video.
• State persistence: Initially, refreshing reset progress. I redesigned data flows with Supabase to persist user history, folders, and settings across sessions.
• AI answer quality: Early versions returned generic responses. By refining system prompts and structuring assistant behavior, I enabled flexible Q&A that supports productivity, context recall, and follow-up learning.
• Monetization flow: Integrated Stripe plans (free + premium trial), ensuring subscription logic works across account states.
• UI/UX: Modeled the sidebar experience after ChatGPT for familiarity, while adding custom features like folder popups and automatic categorization.

Takeaway: This project deepened my skills in full-stack development, AI integration, and SaaS product design. It’s teaching me to handle not just technical challenges but also user experience design, persistence, and monetization—skills critical for building real-world IT/AI platforms.

Outcome

• Status: In Progress
• Deliverables so far:
  • Functional transcript summarization + note generation pipeline.
  • AI assistant chat with context awareness.
  • Sidebar with persistent history and folders.
  • Stripe-based subscription plans and Supabase persistence.

Next Steps:

• Expand transcript fallback pipeline for edge cases.
• Add profile customization (pfp, settings, account switching).
• Improve note quality with timestamped key-point extraction.
• Beta test with real users for feedback.