Scroll Learning

Every day, Gen Z spends hours doomscrolling through memes and cat videos. But what if that same feed could sneak in college-level learning?

Comment

Scroll Learning — Hackathon Submission

Inspiration

We all doomscroll — TikToks, Reels, memes — but rarely does that time translate into learning. Online courses have great content but terrible completion rates. We asked:
What if we could hide learning inside the scroll?

Inspired by Gen Z's content habits, we created a way to turn any course into short, engaging, auto-generated reels — making microlearning as addictive as Instagram.

What it does

Scroll Learning takes any course material (PDFs, slides, notes) and automatically:

  • Breaks it into micro-sized chunks
  • Writes short, Gen Z-style video scripts
  • Generates AI voice narration
  • Adds visuals (either generated or fetched)
  • Stitches it all together
  • Posts it as short-form video content (Reels, TikToks, Shorts)

The result: a personal learning feed disguised as an Instagram page.

Phase 1 focuses on Instagram Reels. Later phases will expand to TikTok, YouTube Shorts, and eventually a native Scroll Learning platform.

How we built it

We built the entire MVP in ~90 minutes using Lovable's AI stack:

  • LLM 1 (Chunker): parses input courses into bite-sized topics
  • LLM 2 (Scriptwriter): writes 30-60 second explainer scripts
  • TTS engine: generates human-like narration automatically
  • Visual layer: retrieves relevant images or generates custom visuals
  • Edge functions: stitch together audio, text, and images into complete video files
  • Instagram API: posts directly to user's learning feed (phase 1)
  • Supabase: database for storing user information

Challenges we ran into

  • Balancing information density vs. engagement time (30-60 seconds is short!)
  • Making scripts sound natural, entertaining, and educational
  • Automated visuals sometimes mismatch the topic (we built fallback safeguards)
  • Instagram API limitations and posting automation hurdles
  • Getting LLM prompt engineering right to avoid "AI textbook voice"

Accomplishments that we're proud of

  • Fully functional MVP built under tight hackathon constraints
  • End-to-end automation: from PDF to posted Reel with zero manual editing
  • Achieved smooth, engaging scripts that actually sound like Gen Z
  • Fully serverless architecture with scalable pipeline design
  • Designed a monetization-ready pathway for creator partnerships and LMS integrations

What we learned

  • The biggest blocker in online learning isn’t content — it’s format.
  • AI excels at turning long-form knowledge into consumable micro-content.
  • EdTech adoption grows faster when it adapts to existing user habits instead of forcing new ones.
  • User attention spans are incredibly short — writing for them requires ruthless editing.

What's next for Scroll Learning

  • ✅ Phase 1: Instagram Reels microlearning (MVP complete)
  • 🔜 Phase 2: Expand to TikTok, YouTube Shorts, and cross-platform content sync
  • 🔜 Phase 3: Build native Scroll Learning app with personalized feeds and progress tracking
  • 🔜 LMS integrations for schools, bootcamps, and coaching programs
  • 🔜 Creator monetization: "Upload & Earn" model for educators and influencers

Built With

  • ai
  • apis
  • lovable
  • supabase
Share this project:

Updates