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

> Invisible, privacy-first AI note taker

- Site: https://bluedothq.com
- Category: AI Meeting Assistant
- Platforms: Google Chrome extension, macOS app, Windows app, iOS app, Android app
- Verdict: **Solid side project** (62/100 vibecodeable)
- Estimated effort: 3-4 weeks of focused coding

## Verdict

Build a personal web-based audio upload and transcription dashboard, but skip writing native cross-platform desktop shells and browser extensions.

Replicating Bluedot's core value for personal use means dropping the native desktop wrappers and bot-free browser extension capture in favor of an audio upload tool that transcribes via Whisper and summarizes via LLMs. Building a robust system-audio capture extension that works across Zoom, Google Meet, and Teams without breaking on browser updates is a massive engineering trap for a solo developer. If you just want automated meeting summaries for personal notes, use existing tools; build this only to learn audio processing pipelines and RAG chat.

### What you can't replicate

- Zero-bot extension capture across 50,000+ corporate environments without breaking permissions
- Audited SOC 2 Type II compliance and enterprise data governance frameworks

## What it does

Bluedot captures, transcribes, and summarises meetings across online platforms and in-person without injecting virtual meeting bots.

### Core features

- Bot-free audio and video recording via Chrome extension or desktop capture
- Multi-language speech-to-text transcription (100+ languages)
- Automated summaries, action items, and key highlights extraction
- Cross-meeting RAG and conversational AI chat interface
- Webhook and Model Context Protocol (MCP) data syncing
- Secure local storage and enterprise data privacy controls

## The business

### Pricing

- Free: $0 — Lifetime limit of 5 meetings
- Basic: $14/mo — The essentials, audio only
- Pro: $20/mo — No limits, with video
- Business: $32/mo — Advanced workflows & controls

### Funding

~$600,000 USD raised.
- Angel Round (May 2024): €560,000 / $600,000
Investors: Google for Startups, Daniel Gross, Greg Jakacki, Tomas James, Bridget Harris

Founded 2021.
Team size: 7 employees.

## The hard parts

- Capturing system audio and browser tab streams cleanly without virtual bot attendance or audio feedback loops
- Managing multi-platform state sync across extension, desktop shell, and web dashboard
- Building a reliable cross-meeting vector search and RAG architecture over large transcripts
- Maintaining low-latency transcription pipelines and structured LLM extraction without data leakage

## How to vibe code Bluedot

### Prerequisites

- Node.js (free): Required for running the Next.js full-stack framework and package manager.
- GitHub (free): Source control and deployment pipeline integration.
- OpenAI API Account (Pay-as-you-go (~$0.006/min)): Provides the Whisper speech-to-text model and GPT-4o for meeting summaries.

### Recommended AI tools

- Claude Code: Best-in-class terminal agent for scaffolding the full-stack app, database schema, and API integrations.
- Cursor: Ideal for fine-tuning UI components, React state management, and debugging transcription upload flows.

### Stack

- Frontend: Next.js with Tailwind CSS and shadcn/ui
- Backend: Next.js Server Actions and API Routes
- Database: Turso (SQLite at the edge with vector support via libSQL)
- Auth: better-auth
- Payments: None (Personal use clone)
- Other: OpenAI API (Whisper STT & GPT-4o), Cloudflare R2 (Audio/Video storage)

### Hosting

- Vercel (Hosting the Next.js frontend and serverless API endpoints): $0-20/mo
- Cloudflare (Storing recorded audio and video files securely in R2 object storage): $0/mo (Free tier)

### Build guide

1. **Project Scaffolding and Database Schema** — Initialize a Next.js project with TypeScript, Tailwind CSS, shadcn/ui, and better-auth connected to Turso database.

```
Create a new Next.js project using App Router, TypeScript, and Tailwind CSS. Configure better-auth with email/password authentication backed by a Turso SQLite database using libSQL. Set up the initial database schema with tables for users, meetings (id, title, audio_url, transcript, summary, action_items, created_at), and chat_messages. Ensure all environment variables are properly structured in .env.example.
```

2. **Audio Upload and Storage Pipeline** — Implement file upload functionality for audio and video files, storing binaries in Cloudflare R2 object storage.

```
Build a secure audio/video upload component and corresponding API route in Next.js that accepts MP3, M4A, and MP4 files up to 500MB. Integrate Cloudflare R2 object storage (via AWS S3 client) to store the raw media files securely and return a signed playback URL. Save the file metadata and storage path linked to the authenticated user's account in the Turso meetings table.
```

3. **Speech-to-Text Transcription Integration** — Integrate OpenAI Whisper API to transcribe uploaded meeting audio files into structured text with timestamps.

```
Write a background processing service using OpenAI's Whisper API (`audio.transcriptions.create`) that downloads the audio file from Cloudflare R2, sends it for transcription with timestamp support, and saves the resulting JSON transcript text into the meeting record in Turso. Handle long audio chunking if file size exceeds API limits.
```

4. **AI Summarization and Action Items Engine** — Create an LLM processing step that analyzes transcripts to generate executive summaries, key decisions, and action items.

```
Implement an LLM summarization pipeline using OpenAI or Anthropic API. When a transcription completes, send the transcript text to the model with a structured prompt instructing it to extract an executive summary, bulleted key points, categorized action items with assignees, and a Q&A section. Store these structured outputs in JSON columns in the Turso database and display them on the meeting details view.
```

5. **Cross-Meeting RAG and AI Chat Interface** — Build a conversational chat interface allowing users to query insights across all past transcribed meetings.

```
Build a multi-meeting AI chat interface where users can ask questions about any past meeting or search across all transcripts. Implement text chunking and vector embeddings using turso/libSQL vector search extensions (or simple SQLite full-text search if vector extension is unavailable) to retrieve relevant meeting context, pass it into an LLM prompt with system instructions, and stream the response back to the frontend UI.
```

### Cost vs paying

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

- AI Coding Assistant subscription: $20
- Domain name (optional): $12
- Total: ~$32 one-time

**Ongoing costs (monthly):**

- OpenAI Whisper & LLM API usage: ~$5-15/mo depending on meeting volume
- Cloudflare R2 & Vercel hosting: $0-5/mo
- Total: ~$5-20/mo

- Paying for the SaaS instead: $20/mo (Pro Plan)
- Build time: 30-40 hours
- AI tool credits: $20 (Claude Code / Cursor Pro)
- Break-even: 1 month of heavy transcription usage

## Sources

- [Bluedot Official Website](https://bluedothq.com)
- [Wikitia - Bluedot Company Profile](https://wikitia.com/wiki/Bluedot)
- [Vestbee - Ukrainian-founded Bluedot raises €560k](https://www.vestbee.com/blog/ukrainian-founded-bluedot-raises-560k)
- [GetLatka - Bluedot Revenue & Valuation Metrics](https://getlatka.com/companies/bluedot)