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

> AI Meeting Notetaker & Automation Platform

- Site: https://meetgeek.ai
- Category: Meeting Intelligence & Productivity
- Platforms: Web app, Desktop app, Mobile app, Browser extension
- Verdict: **Serious undertaking** (45/100 vibecodeable)
- Estimated effort: 6+ weeks of part-time engineering work

## Verdict

Build a single-user personal subset featuring local audio uploads and Whisper transcripts, but expect immense friction handling live browser bot automation.

Replicating MeetGeek end-to-end is a monumental task because of the headless media infrastructure required to auto-join Zoom and Teams calls without getting banned or crashing. While an AI coding agent can easily scaffold a Next.js dashboard, database schemas, and LLM prompting loops, managing real-time audio streams, accurate speaker diarization, and bi-directional API webhook handlers will consume dozens of hours of debugging.

### What you can't replicate

- Enterprise SOC 2 Type II and HIPAA compliance guarantees
- Maintained headless bot fleet capable of bypassing platform auth hurdles on Zoom and Microsoft Teams
- Native integrations across 10,000+ enterprise applications

## What it does

An AI-powered meeting intelligence and automation platform that records, transcribes, summarizes, and analyzes meetings, updating CRMs and triggering agentic workflows.

### Core features

- Multi-modal audio/video recording (bot-based, desktop, mobile, browser extension)
- Automated speech-to-text transcription with speaker diarization
- Custom summary templates and LLM-driven structured extraction
- Cross-meeting conversational AI chat
- Conversation analytics (talk-to-listen ratios, sentiment, keywords)
- Event-driven workflow automation (updating CRMs, tasks, Slack alerts)

## The business

### Pricing

- Basic (Free): $0 — Ideal for occasional meetings.
- Pro: $9.99/mo — For regular meetings with helpful automation.
- Business: $17/mo — Individuals & teams with frequent meetings.
- Enterprise: Custom — For organizations requiring governance and scale.

### Funding

€1.6M raised.
- Seed round (September 2025)
Investors: Early Game Ventures, Inspire Capital

Founded 2020.
Team size: ~12-25 employees.

## The hard parts

- Reliable calendar bot infrastructure that auto-joins Zoom, Google Meet, and Teams without getting blocked
- Real-time audio processing pipelines with precise speaker attribution (diarization) across overlapping speech
- Executing asynchronous background agent loops that parse meeting outputs and sync mutations reliably across external APIs

## How to vibe code MeetGeek

### Prerequisites

- Node.js (free): Required runtime for executing the Next.js frontend and background worker code.
- GitHub (free): Source code repository and CI/CD deployment pipeline.
- OpenAI API Account (pay-as-you-go): Provides the Whisper STT and LLM endpoints required for transcription and summarization.

### Recommended AI tools

- Claude Code: Best-in-class terminal coding agent for scaffolding multi-file full-stack features and fixing complex state bugs.
- Cursor: Ideal AI-native editor for rapidly reviewing UI component diffs and tweaking Tailwind dashboard layouts.

### Stack

- Frontend: Next.js with Tailwind CSS and shadcn/ui components
- Backend: Next.js Server Actions and API routes with Inngest for background jobs
- Database: Turso (SQLite at the edge)
- Auth: better-auth
- Payments: none
- Other: Vercel AI SDK, OpenAI API (Whisper + GPT-4o), Resend

### Hosting

- Vercel (Hosting the Next.js web application and serverless functions): $0-20/mo
- Cloudflare (Storing audio recordings and transcripts in R2 object storage): $0/mo

### Build guide

1. **Project Scaffolding & Database Schema** — Initialize the Next.js project with Tailwind CSS, configure better-auth with SQLite via Turso, and define database models for users, meetings, transcripts, and action items.

```
Scaffold a new Next.js project using TypeScript, Tailwind CSS, and App Router. Set up better-auth for email/password authentication backed by a Turso SQLite database. Create Prisma or Drizzle schema models for Users, Meetings (id, title, duration, audioUrl, createdAt, status), Transcripts (id, meetingId, speaker, text, timestampStart, timestampEnd), and ActionItems (id, meetingId, text, assignee, completed). Add basic layout structure with a responsive sidebar dashboard layout using shadcn/ui primitives.
```

2. **Audio Upload & Transcription Pipeline** — Implement file upload functionality for audio/video files, storing binaries in Cloudflare R2 and processing them through OpenAI's Whisper API.

```
Implement an audio/video upload endpoint in Next.js that accepts files up to 500MB, uploads them directly to Cloudflare R2 object storage, and triggers a background job using Inngest or a server action. The job must send the audio file to the OpenAI Whisper API to retrieve timestamped segments with speaker labels. Save the resulting transcription items into the Transcripts table linked to the meeting record.
```

3. **LLM Summary & Action Item Extraction** — Build the AI summarization engine using the Vercel AI SDK and custom prompt templates to extract meeting summaries and action items.

```
Create a meeting processing workflow using the Vercel AI SDK and Anthropic or OpenAI models. When a transcription is finalized, read all transcript segments and run a structured extraction prompt that returns a JSON object containing: an executive summary, key discussion bullet points, and an array of action items with assignees and due contexts. Save these outputs to the meeting record and display them in a polished summary view on the frontend.
```

4. **Conversational AI Meeting Chat Interface** — Build an interactive chat interface that lets users query past meeting transcripts using vector embeddings or contextual LLM retrieval.

```
Build an AI chat page where users can query their meeting database. Use the Vercel AI SDK useChat hook to power a conversational interface. When a user asks a question (e.g., 'What did we decide about pricing?'), fetch relevant transcript snippets from Turso using full-text search or vector embeddings, inject them as context into the system prompt, and stream the cited response back to the user interface in real time.
```

5. **Analytics Dashboard & Post-Meeting Workflows** — Implement conversation analytics (talk-to-listen ratios) and webhook triggers to push action items into external tools like Slack or Notion.

```
Build a conversation analytics dashboard calculating talk-to-listen ratios, word counts, and keyword frequencies per speaker from the transcript table. Add a workflow configuration screen allowing users to toggle automated triggers when a meeting finishes—such as formatting a Markdown summary and posting it to a designated Slack webhook URL or Notion database via REST API.
```

### Cost vs paying

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

- Custom Domain: $12 one-time
- Total: ~$12 one-time

**Ongoing costs (monthly):**

- Vercel Hobby/Pro Hosting: $0-20/mo
- OpenAI Whisper & LLM API Usage: ~$10-15/mo depending on recording volume
- Total: ~$15-35/mo

- Paying for the SaaS instead: $17/mo (Business tier)
- Build time: 40-50 hours
- AI tool credits: $20/mo (Claude Pro)
- Break-even: Never (purely for learning and personal customization)

## Sources

- [MeetGeek Official Website](https://meetgeek.ai)
- [MeetGeek Pricing & Features Documentation](reports/ab14dc54-a0b3-48d9-8c12-5d4b707155ea/pages/_pricing.md)
- [The Recursive: MeetGeek Secures €1.6M Funding](https://therecursive.com/meetgeek-secures-1-6m-from-early-game-and-inspire-capital-to-scale-its-ai-powered-meeting-assistant-into-a-next-gen-ai-workspace/)