Startup Overview
Problem, solution, target users, current status, and vision.
docs/startup/01-startup-overview.md
A polished reading layer for the startup files, original Google Doc content, and the new internal product documentation. Use this page to orient founders, investors, developers, designers, and future team members.
Markdown files remain the editable internal record. This page organizes them visually and points the team to the right file.
docs/startup/
Primary promise
Raw creator content to polished reels
Current output
9:16 MP4 short videos
Target speed
Under 3 minutes
Render engine
Remotion on AWS Lambda
Transcription
Groq Whisper
Storage policy
Temporary S3 uploads and outputs
Files `01` to `09`, the original Google Doc archive, and the master internal product documentation are all represented here.
Problem, solution, target users, current status, and vision.
docs/startup/01-startup-overview.md
Video type catalog, inputs, outputs, future video type ideas, and format rules.
docs/startup/02-product-and-features.md
Stack, render pipeline, deployment flow, costs, and Supabase tables.
docs/startup/03-technical-architecture.md
Video type creation, Remotion rules, naming, testing, and quality standards.
docs/startup/04-video-type-rules.md
English and Hinglish caption rules, paused translation scope, and provider policy.
docs/startup/05-subtitle-language-rules.md
Asset storage, S3 lifecycle, CDN rules, and Vercel deployment constraints.
docs/startup/06-assets-and-s3.md
Common render, subtitle, asset, and deployment issues with fixes.
docs/startup/07-known-issues-and-fixes.md
Immediate, short-term, and future product direction.
docs/startup/08-roadmap.md
YC advice, decisions log, metrics, demo notes, and competitive landscape.
docs/startup/09-yc-investor-notes.md
Original combined startup documentation imported from Google Docs.
docs/startup/GOOGLE_DOC_CONTENT.md
Reorganized professional master doc with video types, trackers, risks, APIs, and roadmap.
docs/startup/ITNAVIDEO_INTERNAL_PRODUCT_DOCUMENTATION.md
The docs preserve both the current production context and the founder notes. Custom AI Reel is kept visible, but its production status should be confirmed because documentation sources differ.
Current project context says 6 production video types. The Google Docs source lists 7 including Custom AI Reel. The master documentation marks this as a confirmation item.
| Video type | Purpose | Input | Output |
|---|---|---|---|
| Auto Caption Video | Captioned reels from uploaded videos | Video with speech | Captioned MP4 |
| Dynamic Explainer Video | Text cards, icons, highlights, motion | Audio/script/video | Explainer reel |
| Compare Explainer Video | Side-by-side comparison with presenter elements | Audio + 2 sides | VS reel |
| Auto Draw Explainer Video | Notebook or whiteboard style teaching video | Audio/script | Animated notes reel |
| Long Video Promo | Promo reel for long videos | Clip + thumbnail + title | Promo short |
| Background Replace Video | Clean vertical video with changed background | Video + background | Background replaced reel |
| Custom AI Reel | Prompt-led custom reel workflow | Prompt + optional media | Needs status confirmation |
The core product flow from upload to final MP4.
Creator selects a video type and uploads the required media.
API validates file type, user, credits, and required inputs.
Uploads go to S3 through temporary secure URLs.
FFmpeg extracts audio where needed; Groq Whisper returns transcript and timestamps.
Local planner or approved AI planner builds scenes, captions, and render props.
Remotion Lambda creates the 9:16 MP4.
Job status returns progress, final URL, download, and history entry.
| Layer | Technology | Role |
|---|---|---|
| Frontend | Next.js App Router, React, Tailwind CSS | Dashboard, pages, API entry points |
| Auth and data | Supabase | Accounts, render history, app data, credits |
| Transcription | Groq Whisper | Audio to transcript and word-level timestamps |
| Planning | Local planners + Gemini for Auto Draw | Scenes, captions, timelines, visual instructions |
| Rendering | Remotion Lambda | Animated compositions and final video render |
| Media processing | FFmpeg | Audio extraction, trim, convert, compress, screenshot, merge |
| Storage | AWS S3 | Temporary uploads, temporary render files, final outputs |
| Payments | Razorpay | Plans, credits, paid user flow |
The master documentation explains the product. The docs hub also shows shapes and examples so developers can understand how captions, settings, and render props fit together.
Examples stay close to the actual render pipeline.
Callouts and tables make the system easy to explain.
{
"compositionId": "AUTO-CAPTION-REEL",
"durationSeconds": 60,
"mediaSrc": "https://signed-s3-url.example/input.mp4",
"captions": [
{
"start": 0.4,
"end": 1.8,
"text": "create better reels",
"words": [
{ "word": "create", "start": 0.4, "end": 0.7 },
{ "word": "better", "start": 0.7, "end": 1.1 },
{ "word": "reels", "start": 1.1, "end": 1.8 }
]
}
],
"captionStyle": "studioClean",
"settings": {
"position": "bottom-safe-area",
"font": "Inter",
"highlight": "Yellow"
}
}The new master doc adds the working trackers needed for a startup team.
Master doc
Tracks total video types, status, owner, priority, issues, planned improvements, and testing state.
Operations
Tracks ID, date, module, severity, root cause, proposed fix, owner, and resolution date.
Product
Tracks module, current state, suggested improvement, expected impact, priority, and status.
Founder notes
Tracks decisions such as provider changes, video type simplification, and language scope.
Vercel for frontend/API and Remotion Lambda for render engine.
Supabase handles auth, history, credits, app data, and user state.
Remotion compositions render serverlessly with S3 media URLs.
Private files, protected APIs, limited URLs, safe logs, and credit protection.