# How to Vibe Code Your Own Subclip (and Stop Paying for It)

> Prepare Videos for Global Reach- Viral Captions, Translation & AI clipping

- Site: https://subclip.app
- Category: Video Production & AI
- Verdict: **Serious undertaking** (38/100 vibecodeable)
- Estimated effort: 6-8 weeks of part-time work

## Verdict

Build a simple transcript editor wrapper rather than paying, but a full Subclip clone requires months of heavy pipeline engineering.

Replicating Subclip as a personal tool is an ambitious multi-week engineering project. While you can easily stitch together frontend video trimming and OpenAI Whisper transcripts using modern AI tools, orchestrating background FFmpeg rendering jobs, word-level caption canvas animations, and reliable multi-step AI video dubbing pipelines will expose you to severe timeout hurdles, storage bills, and complex async state management.

### What you can't replicate

- Subclip's fine-tuned viral scoring and automated hook detection heuristics
- Their native macOS and iOS companion apps optimized for hardware-accelerated rendering
- Production-grade infrastructure handling thousands of concurrent high-resolution video exports

## What it does

An AI-native video post-production and repurposing platform that automates clipping long-form content, generating dynamic captions, dubbing into 21+ languages with voice cloning, and transcript-based editing.

### Core features

- AI video clipping (detecting punchlines and emotional shifts)
- Dynamic auto-captions with kinetic styling and face-tracking
- AI video dubbing and voice cloning in 21+ languages
- Transcript-based video editing (Google Doc style trimming)
- NLE timeline exports (Premiere, Final Cut, DaVinci Resolve)
- Social media publishing and scheduling
- Audio enhancement and noise removal

## The business

### Pricing

- Trial: $4.99 one-time
- Creator: $14/mo
- Pay as You Go: $99 one-time
Team size: Small indie team.

## The hard parts

- Synchronizing word-level timestamps with animated canvas/CSS caption overlays
- Managing heavy asynchronous media rendering pipelines (FFmpeg) without hitting serverless timeouts
- Orchestrating multi-model AI workflows (Whisper transcription, LLM script analysis, ElevenLabs voice cloning)
- Handling multi-gigabyte video uploads, signed storage URLs, and cloud egress cost optimization

## How to vibe code Subclip

### Prerequisites

- Node.js (free): Required for running the full-stack web application framework.
- GitHub (free): Repository hosting and CI/CD deployment pipelines.
- FFmpeg (free): Core media manipulation binary required locally and on the server for cutting, caption burning, and muxing.

### Recommended AI tools

- Claude Code: Best-in-class agentic coding tool for scaffolding full-stack architectures, writing complex FFmpeg wrapper scripts, and debugging async pipelines.
- Cursor: Ideal for fine-tuning the interactive video editor UI and timeline components with direct diff reviews.

### Stack

- Frontend: Next.js
- Backend: Next.js API routes + Trigger.dev
- Database: Turso
- Auth: better-auth
- Payments: none
- Other: Vercel AI SDK, OpenAI API (Whisper), ElevenLabs, FFmpeg

### Hosting

- Fly.io (Running long-lived Docker containers equipped with FFmpeg for heavy video rendering and transcoding jobs.): $5-15/mo
- Cloudflare R2 (Storing multi-gigabyte source videos and exported clips with zero egress fees.): $0-5/mo

### Build guide

1. **Scaffold Project and Database Schema** — Initialize the Next.js full-stack application with Tailwind CSS, configure Turso for SQLite storage, and implement better-auth for single-user/personal authentication.

```
Initialize a new Next.js project with TypeScript and Tailwind CSS. Set up better-auth connected to a Turso database via libSQL. Create database tables for users, projects, videos (storing R2 object keys, duration, and status), transcripts (storing JSON word-level timestamps), and clips. Write a clean landing page and a dashboard layout with sidebar navigation for Projects, Editor, Dubbing, and Settings. Ensure all environment variables are correctly configured and validation uses Zod schemas.
```

2. **Video Upload & Cloudflare R2 Integration** — Implement direct-to-storage or server-mediated video uploads to Cloudflare R2 with signed upload/download URLs and progress tracking.

```
Build a secure video upload workflow in Next.js using Cloudflare R2 via AWS SDK v3. Create an API route that generates presigned PUT URLs for client-side uploads to avoid server memory overload on large video files. Implement a dropzone component with upload progress indication. Upon successful upload, create a video record in the Turso database and trigger a background verification job to extract basic metadata (resolution, codec, duration) using fluent-ffmpeg.
```

3. **AI Transcription & Word-Level Timestamp Pipeline** — Integrate OpenAI Whisper API to transcribe uploaded videos and parse word-level timestamps for subtitle synchronization.

```
Create a transcription background worker using Trigger.dev that downloads the video audio track, sends it to the OpenAI Whisper API with response_format='verbose_json' and timestamp_granularities=['word'], and parses the resulting word timestamps. Save the structured transcript JSON into the database. Build a transcript viewing component in React that highlights words as a mock player ticks along, allowing text selection and deletion.
```

4. **Transcript-Based Video Editor & FFmpeg Trimming** — Build a Google-Doc style text editor linked to video playback, enabling rough cuts by deleting transcript sections which then trigger FFmpeg re-encoding.

```
Build an interactive transcript editor component where users can select text spans and delete them to remove silences or mistakes. When changes are saved, generate an edit decision list (EDL) mapping retained time segments. Write a server-side FFmpeg processing script that takes the EDL, slices the source video into segments, concatenates them cleanly using the concat demuxer, and exports a new edited video file saved back to R2.
```

5. **AI Viral Clip Extraction** — Integrate an LLM prompt pipeline to analyze transcripts, identify high-engagement segments, and automatically slice clips into vertical 9:16 aspect ratios.

```
Implement an AI clipping service using the Vercel AI SDK and Anthropic Claude. Send the full transcript text with a system prompt instructing the model to identify 3 to 5 high-energy hook moments with start and end timestamps, accompanied by engagement rationale and viral scores. Build a UI dashboard displaying these suggested clips, allowing users to click 'Export Clip' which triggers an FFmpeg job to crop the video to 9:16 vertical format and burn or overlay subtitles.
```

6. **Dynamic Auto-Captions Burn-in** — Render time-synced animated subtitles with styling templates (bold, one-word kinetic formats) burned directly onto video frames using FFmpeg drawtext or ASS filters.

```
Build a subtitle styling customization panel (selecting fonts, colors, and layouts like One-Word or Bold Kinetic). Write a Python or Node utility that converts word-level transcript timestamps into an ASS (Advanced Sub Station) subtitle file with precise styling tags. Create an FFmpeg processing job that burns the ASS subtitles directly onto the video frames during export, ensuring accurate synchronization and zero font-rendering discrepancies.
```

### Cost vs paying

**Starting costs (one-time):**

- Claude Pro / Cursor subscription: $20
- Total: ~$20 one-time

**Ongoing costs (monthly):**

- Fly.io compute hosting: $8/mo
- Cloudflare R2 storage & egress: $2/mo
- OpenAI Whisper & LLM API usage: $5/mo
- Total: ~$15/mo

- Paying for the SaaS instead: $14/mo
- Build time: 45-60 hours
- AI tool credits: $20 (Claude Pro)
- Break-even: Never (built for learning and custom control)

## Sources

- [Subclip Product Overview & Features - IndieHunt](https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQFMZOcultLUozUSHtgq82tBLWnE5h9EnS21koEfxf320Kw-u-zBir_xOWfo6jelwEdwT68U-tsfTS-BU0bjD8FW3GNEeqlshFPrW3gEe-GFf1B6NgKm_IxYpPA=)
- [What Is Subclip? Platform Overview](https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQFq6iUvaBwLxbyKzZqCTlfGvVJvdytUX9ei4hCZL-22Dis_ObcuZKQeHQa93S9JaHhuPY_jxu9oLZPI2GG4w2mJfkXMeJ6ehk3DA7ydii8OPUVzaFAgk8uc_dMqQfib3aZu)
- [Subclip Trustpilot Reviews & Company Profile](https://vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIYQH58BSRTilfx_hpb-o_b0xI4FNm0YeSKfteYIqLpmie-uhtDT_-cDJQiBDmr4Tm-69vm7bxThacsx72I9zxkGi13F0EEwyHSJY2ljtA1_VA1jl85xekmLFeDQZTtM6tv-lDoFg=)