How to vibe codeMaze
User Research and Testing Platform
maze.co ↗Product Discovery & User Research SaaS
The verdict: can you vibe code Maze?
Build a personal-use subset of the study builder, survey logic, and AI synthesis dashboard — but keep paying for Maze if you need the 6M+ participant recruitment panel.
Replicating Maze as a solo developer means building several complex subsystems: a visual study editor, a prototype testing engine that parses Figma layouts and hot zones, an event-tracking ingestion pipeline for session analytics, and an AI moderation engine with voice streaming. While the study builder and reporting UI can be vibecoded swiftly using Next.js and Vercel AI SDK, orchestrating low-latency conversational AI and real-time interaction heatmaps is a serious multi-month engineering lift. The true moat—access to millions of vetted participants—is completely impossible to clone.
Estimated effort: 3-4 months of part-time work
What you can't replicate
- Global participant panel of 6M+ vetted B2B and consumer testers
- Enterprise SOC 2 Type II compliance and security governance
- Native mobile companion app recording infrastructure for iOS and Android
Founded
2018
Raised
$60M
Team
150-170
Cheapest paid tier
$0/mo
What Maze does
Connect everyone to users with our end-to-end research platform for prototype testing, surveys, tree testing, card sorting, and AI-moderated interviews.
Core features
- Study Builder with various question and task types (surveys, card sort, tree test, prototype test)
- Figma prototype integration via embedding and hotspot parsing
- Unmoderated session recording (clicks, mouse tracks, time-on-task, drop-off paths)
- AI Moderator for autonomous conversational interviews with audio streams
- Participant management and screeners
- Automated reports with video clip generation and synthesis
- MCP (Model Context Protocol) server for external AI tool integration
- In-product prompt snippet embedding for live web apps
The business
Pricing
- Free$0/mo
- Starter$99/mo
- EnterpriseCustom
Funding
$60M from Amplify Partners, Felicis Ventures, Emergence Capital, Atlassian Ventures, HubSpot Ventures, Zoom Ventures, Seedcamp
Pay vs build, cumulative
Break-even at month 1 — after that, every month is money kept.
The hard parts of vibe coding Maze
- Figma design API synchronization and real-time hotspot hit-testing across scaled artboards
- Event streaming ingestion pipeline capturing high-frequency click/cursor metrics from concurrent unmoderated sessions
- Low-latency conversational AI orchestration loop managing voice transcription, goal evaluation, and audio generation
- GDPR-compliant video storage, masking, and automated transcription indexing
How to vibecode Maze
Prerequisites
Node.jsfree
Runtime for running the full-stack web application
GitHubfree
Source code repository and CI/CD deployment connection
OpenAI API Keypay-per-use
Powers study generation and transcript synthesis
AI coding tools
Recommended stack
| Frontend | Next.js |
|---|---|
| Backend | Next.js API Routes |
| Database | Neon |
| Auth | better-auth |
| Payments | none |
| Other | Vercel AI SDK, Resend, Deepgram, ElevenLabs |
Build guide
01Project Scaffolding and Database Schema
Initialize the Next.js app with Tailwind CSS, configure Neon Postgres connection, and set up better-auth tables.
Initialize a new Next.js project with TypeScript, Tailwind CSS, and App Router. Set up Drizzle ORM connected to Neon Postgres. Create database schemas for users, workspaces, studies (title, description, status), blocks (question, card_sort, tree_test, prototype_test), and responses. Implement better-auth for email/password authentication with secure session handling. Add a clean dashboard shell with sidebar navigation for studies, participants, and reports.02Study Builder & Block Editor
Build a modular study builder interface allowing creators to add, reorder, and configure different research block types.
Build a drag-and-drop study builder interface in Next.js using shadcn/ui components. Allow creators to add blocks: Welcome screen, Multiple Choice Survey, Open Question, Card Sorting, Tree Testing, and Prototype Testing. Each block should have a configuration panel for questions, options, and required flags. Persist block states instantly to Neon Postgres via server actions.03Figma Prototype Integration & Hotspot Testing
Implement prototype test rendering by embedding Figma frames and defining clickable hotspots to record interaction paths.
Create a prototype testing participant view that allows embedding Figma frames or uploaded images. Build an interactive editor mode where researchers can draw bounding-box hotspots over image areas and designate target destination screens. In participant test mode, record click coordinates, misclicks, success/failure status, and time-on-task, saving metrics to the database.04AI Moderator & Audio Interview Simulator
Integrate Vercel AI SDK and ElevenLabs/Deepgram for an automated conversational interview mode driven by study goals.
Build an AI-moderated interview screen using Vercel AI SDK and the OpenAI API. Allow researchers to input study goals and a discussion guide. Implement a live voice interface where participants speak, audio is transcribed via Deepgram, processed by the LLM agent to generate goal-aligned follow-up probes, and spoken back using ElevenLabs text-to-speech. Save full transcripts and audio clips to the database.05Analytics Dashboard & Automated Reporting
Aggregate session metrics, generate quantitative usability charts, and synthesize qualitative summaries using LLMs.
Build an automated reporting dashboard for completed studies. Compute and display quantitative metrics including success rates, misclick rates, average time-on-task, and survey distributions using Recharts. Integrate an AI synthesis button that calls the LLM with all participant transcripts and open responses to generate grouped thematic insights and extract key video/text quote highlights.
Cost vs paying for Maze
What will you build it with?
Starting total with Claude Code$0 one-time
Starting costs (one-time)
- Domain name$12 one-time
- AI API starting credits$20 one-time
Total~$32 one-time
Ongoing costs (monthly)
- Vercel Hosting$0-20/mo
- Neon Postgres$0/mo
- OpenAI & ElevenLabs API usage~$10/mo
Total~$10-30/mo
Paying for Maze
$99/mo (Starter) to $1,000+/mo (Enterprise)
Your time to build
60-80 hours
AI tool credits
$20 (Claude Pro) + ~$15 API credits
Break-even
Immediate if building for personal/internal portfolio use; impossible to break even on utility compared to paying for panel recruitment.
Vibe code Maze: FAQ
- Can you vibe code Maze yourself?
- Serious undertaking — 45/100 vibecodeable. Build a personal-use subset of the study builder, survey logic, and AI synthesis dashboard — but keep paying for Maze if you need the 6M+ participant recruitment panel.
- How long does it take to vibe code Maze?
- 3-4 months of part-time work — roughly 60-80 hours of hands-on time with an AI coding agent.
- How do you build your own Maze?
- Scoped to personal use: Next.js on the front, Next.js API Routes behind it, Neon for data. Follow the 5-step build guide on this page — each step has a paste-ready prompt for an AI coding agent.
- How do you code your own Maze without being an expert?
- Use an AI coding tool (Claude Code or Cursor) and work in small steps: scaffold, data model, core screens, then deploy. Realistic effort: 3-4 months of part-time work. The prompts on this page are written so the AI does the heavy lifting.
- How much does it cost to vibe code Maze instead of paying?
- About ~$32 one-time to start and ~$10-30/mo to run, versus $99/mo (Starter) to $1,000+/mo (Enterprise) for Maze. Break-even: Immediate if building for personal/internal portfolio use; impossible to break even on utility compared to paying for panel recruitment..
- What stack should you use to vibe code Maze?
- Next.js; Next.js API Routes; Neon; plus Vercel AI SDK, Resend, Deepgram, ElevenLabs.