Scroll-Stack

The description states that Scroll-Stack converts PDF books into manga-style editions and vertical reels. It runs a pipeline from input PDF to:

• Versioned context pack (source truth)
• Manga plan + page scripts
• Thumbnails and composition
• Rendered pages and manifest
• ReelSpec for vertical cuts

The system produces:

• A manga reader interface (/manga/{edition})
• A library of created editions (/library)
• Vertical reels playable through Remotion engine (Reels)

It uses FastAPI, Next.js, Python, React, and Remotion. The authors claim to have built a deterministic pipeline where:

• Editions are immutable
• Rejected attempts are stored with costs
• Panels trace back to source units in the PDF
• No embedded text in generated images (lettering comes from renderer)

Evidence Self-reported product description.

Inference This is a proof-of-concept tool built for a hackathon, not yet proven in production or at scale.

The authors state that Scroll-Stack was inspired by the problem of people abandoning books due to "cold start" — needing to rebuild world context every time they open a book. They contrast this with manga and reels which provide immediate immersion.

They claim:

• This is not an educational tool or study aid
• It's entertainment made from good books
• The goal is to make reading feel like scrolling through content you always finish

Evidence Self-reported positioning narrative.

Inference The positioning reflects a shift from "how do we get people to read more" to "what if the book you never finished came to you as the thing you always finish?" This suggests an attempt to reframe reading as a form of entertainment consumption rather than discipline-based activity.

The description does not explicitly name target customers or define ideal customer profiles (ICP). However, it implies:

• People who abandon books after page 40
• Users who spend significant time on reels/social media
• Readers who enjoy manga and want similar experiences from books
• Those seeking a more engaging way to consume long-form content

Evidence Implicit in the inspiration story and product claims.

Inference The target appears to be casual readers or those looking for alternative formats to traditional reading, particularly those who struggle with sustained attention on text-based media.

There is no evidence of pricing structure, monetization strategy, or business model in the description. The authors state:

• No quizzes, streaks, progress bars
• Not a study tool
• Focus on entertainment that happens to be made of good books

Evidence Not evidenced.

Inference The product seems to be presented as a service with no clear commercial model described beyond its hackathon prototype.

The system architecture includes:

• FastAPI + Pydantic control plane
• MongoDB (Beanie) for durable state
• Celery + Redis for workflow management
• Next.js / React frontend
• Remotion for video rendering

Key technical features mentioned:

• Deterministic generation pipeline
• Immutable editions with lineage tracking
• Rejection visibility and cost transparency
• Cross-language contract seam via Pydantic models
• Parallel development using worktrees and PR-only integration

Evidence Self-reported technical details.

Inference The team shows strong engineering discipline, especially around determinism, parallelism, and contract-first design. These are signals of a capable engineering team but do not indicate product-market fit or scalability beyond demo-level usage.

The description states:

• This is a hackathon project (OpenAI 2026)
• No revenue, customers, or adoption data
• Demo edition produced from pages 1–15 of a real PDF with zero new text tokens
• Reused accepted upstream artifacts
• One rejected panel was retried once; no accepted panel regenerated

Evidence Self-reported demo execution.

Inference The product exists only as a prototype. There is no evidence of traction, user feedback, or commercial viability beyond the authors' own demonstration.

The description does not mention competitors or market context. It focuses on the internal problem (cold start) and solution (manga/reels format), without reference to existing tools or platforms that might offer similar functionality.

Evidence Not evidenced.

Inference Without knowing the competitive landscape, it's impossible to assess whether this addresses a gap in the market or duplicates existing offerings.

Key risks identified from the description:

• No revenue, customers, or traction data — all self-reported
• Product is limited to a hackathon demo with no production use case
• Heavy reliance on AI image generation (which may be expensive and unpredictable)
• Limited scope of functionality (only 15 pages in demo)
• No indication of scalability or infrastructure for real-world deployment

Evidence Self-reported claims, lack of external data.

Inference The project is unproven in terms of commercial viability. It’s unclear if the team can scale beyond the demo or if there’s sufficient demand for such a service.

1. What specific user problems are you solving, and how do you know they exist?
2. How will you monetize this product? Is there any revenue model being tested?
3. Can you demonstrate any real-world usage or feedback from users beyond the demo?
4. What is your plan for scaling beyond the current hackathon prototype?
5. Are you planning to build a full production version, and what are the key milestones?
6. How do you intend to handle content licensing or copyright issues with books?
7. Have you considered how to make this accessible to people without high-end devices?