01 RABBITHOLE

DECEMBER 2025 • PRODUCT DEEP DIVE

RABBITHOLE

Curing "TikTok Brain" with AI-generated, swipeable knowledge paths.

Launch App

Next.js OpenAI Vercel AI SDK Capacitor Supabase

Use arrow keys or buttons below to navigate

02 The "TikTok Brain" Problem

PART I: MARKET NARRATIVE 02 / 11

The "TikTok Brain" Problem

User attention spans have dropped to ~47 seconds. The "variable reward schedule" of feeds makes traditional reading feel "insufferably boring."

"Popcorn Brain"

Deficits in inhibitory control; users physically struggle to stop switching tasks.

The "Tsundoku" Guilt

Buying books for the aesthetic but failing to read them creates anxiety and loss of identity.

03 The Billion-Dollar Gap

PART I: MARKET NARRATIVE 03 / 11

The Billion-Dollar Gap

While print sales fall, active learning is migrating to audio and micro-formats. Users want to learn but need lower friction.

TAM

600M+

Podcast listeners + micro-learning apps.

TOM

~50M

Smartphone users seeking self-improvement.

SOM

500k

$30M-$50M ARR opportunity.

04 Project note 4

PART II: PRODUCT DEFINITION 04 / 11

Formal Problem Statement

"For knowledge-seeking users, current SFV consumption leads to 'brain rot' because the dopamine loop creates a 'scan-and-shift' state that makes deep reading physically difficult."

Functional Job:
Utilize 5-15 min gaps for discovery.

Emotional Job:
Convert "wasted time" guilt into curiosity.

05 Solution Options

PART III: SOLUTION SPACE 05 / 11

Solution Options

Option	Pros / Cons	Verdict
Micro-Chapters	High depth but high friction (cognitive load).	Discarded
Audio-First	Lowest friction but passive retention.	Postponed (V2)
Smart Card Feed	Matches swipe habit; granular; skimmable.	Selected (V1)

06 The V1 Solution

PART III: SOLUTION SPACE 06 / 11

The V1 Solution

Elevator Pitch:
A swipeable, card-based feed of AI-generated facts that can be explored via tap-to-deepen "holes" within a 5–15 minute session.

Core Mechanics

Infinite Topic Feed
Micro-Holes (Tap to spawn sub-feed)
"Facts Collected" Counter

Success KPIs

Median cards/session > 10
Session felt "productive"

07 V1 Core Flows Demo

PART IV: UX ARTIFACTS 07 / 11

V1 Core Flows Demo

Watch Video

Topic Selection → Infinite Feed → Rabbit Hole Drill Down → Session Summary

08 Friction Analysis

PART V: RESEARCH APPENDIX 08 / 11

Friction Analysis

Frustration	User Description	Validating Source
"The Glaze Over"	Eyes moving without comprehension.	SDSU Eye-tracking Study
"The Itch"	Discomfort when not checking phone.	Psychological Bulletin 2025
"The Audit"	Listening but retaining little.	Active vs Passive Learning

Primary Sources:

09 The "Infinite Wikipedia"

PART VI: TECHNICAL SOLUTION 09 / 11

The "Infinite Wikipedia"

Orchestrating AI agents to synthesize knowledge feeds in real-time.

Intent Analysis: Agents analyze query depth.
Content Synthesis: Parallel LLMs draft content.
Aggressive Streaming: Vercel AI SDK for low latency.

export async function POST(req) {
  const { messages } = await req.json();

  // Parallel agent execution
  const response = await openai.createChatCompletion({
    model: 'gpt-4',
    stream: true,
    messages
  });

  return OpenAIStream(response);
}

10 Mobile Adaptation

PART VI: TECHNICAL SOLUTION 10 / 11

Mobile Adaptation

Built with Capacitor to bring the Next.js web app to Android with native interactions.

Native Gestures

Swipe-to-save, tap-to-drill-down optimized for thumbs.

Offline Caching

Local storage for previously visited "holes".

Share Intent

Deep linking into specific knowledge stacks.

11 FUTURE ROADMAP

FUTURE ROADMAP

Exploring multi-modal learning paths with Flux for images and audio summaries.

Launch Live Demo

Rabbithole

Project presentation