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

> Never Take Call Notes Again

- Site: https://superscribe.io
- Category: Voice AI & Productivity Suite
- Platforms: macOS app, Windows app, iOS app, Web app
- Verdict: **Serious undertaking** (35/100 vibecodeable)
- Estimated effort: 3 to 4 months of part-time work

## Verdict

Build a personal web dashboard and a desktop transcription wrapper, but keep paying for the iOS call capture and native system injection.

The web interface, OpenAI summary pipelines, and desktop audio recording loop are straightforward side project material. However, intercepting inbound and outbound cellular calls on an unjailbroken iPhone while keeping your carrier number requires complex VoIP routing gateways and carrier trunks. Furthermore, building a cross-platform native desktop tool that grabs global hotkeys, streams audio, and safely injects text into arbitrary third-party applications without triggering OS security blocks involves deep platform-specific systems programming.

### What you can't replicate

- Carrier-level iOS call interception infrastructure
- Zero-latency native OS accessibility text injection across arbitrary third-party windows

## What it does

Voice-first productivity and workflow suite capturing mobile phone calls with live transcription, summary, and action items alongside desktop dictation and automatic project time tracking.

### Core features

- Global hotkey-activated real-time voice dictation streaming into any focused app
- Automatic project time tracking linked to dictation and active window context
- iOS call capture and interception preserving existing phone numbers
- Real-time audio streaming websockets connected to voice transcription APIs
- LLM post-call processing extracting action items, CRM notes, and follow-up emails
- Web dashboard for searchable call history, task lists, and project templates

## The business

### Pricing

- Business Voice: $38 / seat / mo
- Travel Pass: $12

Founded 2024.
Team size: Solo developer.

## The hard parts

- iOS telephony interception and carrier-level call routing without violating Apple background audio sandboxing rules
- Cross-platform desktop accessibility hooks for character-by-character text injection into arbitrary third-party native apps
- Ultra-low-latency bi-directional WebSocket audio streaming to ElevenLabs or custom STT engines without jitter
- Heuristic active-window context detection to automatically categorize time tracking sessions into distinct user projects

## How to vibe code Superscribe

### Prerequisites

- Node.js (free): Runtime for building the Next.js web dashboard and backend APIs
- GitHub (free): Source code repository and CI deployment

### Recommended AI tools

- Claude Code: Autonomous multi-file agentic coding tool for scaffolding backend schemas, UI components, and API integration loops
- Cursor: AI-native code editor for fine-tuning desktop dictation hooks and UI components

### Stack

- Frontend: Next.js
- Backend: Node.js API routes
- Database: Turso
- Auth: better-auth
- Payments: none
- Other: ElevenLabs API, OpenAI API, Vercel AI SDK

### Hosting

- Vercel (Hosting the Next.js web dashboard and backend serverless endpoints): $0-20/mo
- Turso (Serverless edge SQLite database for user notes, tasks, and project logs): $0/mo

### Build guide

1. **Project Scaffolding & Database Schema** — Initialize the Next.js project with Tailwind CSS, configure Turso database with Drizzle ORM, and set up user authentication.

```
Create a new Next.js 16 project with TypeScript and Tailwind CSS v4. Set up Turso for database storage using Drizzle ORM. Define database schemas for users, projects, call transcripts, summaries, and time tracking entries. Implement better-auth for secure user login with email and password. Ensure all database migration scripts are written and verified. Verify that the login screen renders cleanly and successfully authenticates a test user into the Turso database.
```

2. **Web Dashboard & Call History UI** — Build the main dashboard for viewing transcript archives, summaries, tasks, and project allocations.

```
Build a responsive web dashboard in Next.js using Tailwind CSS that mimics a clean productivity app interface. Create views for: 1) Call history list with search and filtering by project, 2) Detailed view showing transcript text, AI-generated summary, extracted action items, and follow-up email drafts, 3) Project management view for assigning time-tracking rules. Implement server actions for CRUD operations on transcripts and projects backed by Turso.
```

3. **AI Post-Call Processing Pipeline** — Integrate OpenAI and Vercel AI SDK to automatically process raw transcripts into action items, CRM notes, and follow-up drafts.

```
Implement an AI processing pipeline using the Vercel AI SDK and OpenAI API. Create a backend API route that accepts a raw transcript text string, sends it to GPT-4o with a structured prompt, and extracts JSON containing: a concise summary, key decisions, bulleted action items with assignees, and a draft follow-up email. Store the structured output in the Turso database linked to the specific call record and display it dynamically on the dashboard view.
```

4. **Desktop Dictation & Audio Streaming Integration** — Connect ElevenLabs API for real-time speech-to-text audio stream handling and test dictation flows.

```
Build a web-based audio recording and streaming component that captures microphone input via the browser MediaRecorder API. Implement a WebSocket client handler that streams audio chunks in real time to the ElevenLabs speech-to-text API (or a simulated endpoint if testing locally) and renders the live transcribed text in a streaming text box. Handle audio stream cleanup, silence detection, and error states gracefully.
```

5. **Time Tracking & Project Context Engine** — Add logic to associate dictation and call sessions with active projects for automated time sheets.

```
Implement a time-tracking module within the Next.js backend and dashboard. Allow users to define active projects with keyword tags and target applications. When a dictation or call session is logged, automatically match metadata or user selection to a project, track elapsed duration, and aggregate billable hours on a weekly summary view with charts.
```

6. **Polish, Error Handling, and Deployment** — Add Sentry error tracking, optimize responsive layouts, and deploy the application to Vercel.

```
Perform a comprehensive polish pass on the Next.js application. Integrate Sentry for error tracking across client and server routes. Add loading skeletons, empty states for transcripts and projects, and toast notifications for user actions. Verify all environment variables for Turso, better-auth, OpenAI, and ElevenLabs are correctly configured. Prepare the build for deployment on Vercel and verify production health.
```

### Cost vs paying

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

- Domain name registration: $12 one-time
- Total: $12 one-time

**Ongoing costs (monthly):**

- Vercel hosting: $0-5/mo
- Turso database: $0/mo
- OpenAI & ElevenLabs API usage: $5-15/mo
- Total: ~$10-20/mo

- Paying for the SaaS instead: $38/seat/mo
- Build time: 35-45 hours
- AI tool credits: $20 (Claude Pro)
- Break-even: 1 month

## Sources

- [Superscribe Official Website](https://superscribe.io)
- [Superscribe API Documentation](https://superscribe.io/api)