The most common question people ask about vibe coding is: "Can you actually build something real with this?" The answer is yes — and people already are. SaaS tools collecting monthly revenue, marketplaces connecting buyers and sellers, internal tools replacing spreadsheet workflows, and mobile apps with thousands of downloads.
This article profiles 10 representative projects built with vibe coding tools. These profiles are composites based on publicly shared projects in the vibe coding community — the kinds of stories you will find on Indie Hackers, X/Twitter build-in-public threads, and Product Hunt launches. We have kept details realistic and honest, including the failures and limitations, because that is more useful than hype.
Note: These are representative examples inspired by real patterns we have observed in the vibe coding community. Individual details are illustrative, not biographical.
1. FeedbackOwl — Customer Feedback SaaS
Builder background: Product manager with 8 years of experience, no coding background
Tools used: Lovable (initial build), then migrated to Cursor for customization
Stack: Next.js, Supabase, Vercel, Lemon Squeezy
Timeline: 3 weeks from idea to launch
Status: $2,400 MRR after 6 months
FeedbackOwl is an embeddable widget that lets SaaS companies collect user feedback directly in their app. The builder described the entire product to Lovable in a detailed prompt and had a working prototype in one afternoon. She spent two more weeks refining the UI, adding Supabase for data storage, and integrating Lemon Squeezy for subscriptions.
What went well: The initial prototype impressed early beta users enough to validate the idea before significant time investment. Lovable generated a clean React codebase that was portable when she later switched to Cursor for more control.
What was hard: The embeddable widget required understanding iframes and cross-origin communication — concepts the AI could implement but that were difficult to debug when things went wrong. She also underestimated the complexity of email notifications, eventually adding Resend for transactional emails.
2. LocalMenu — Restaurant Ordering Platform
Builder background: Restaurant owner with no technical background
Tools used: Bolt.new (full build)
Stack: React, Supabase, Netlify
Timeline: 1 week for MVP
Status: Used by 3 local restaurants, not monetized (personal use)
A restaurant owner frustrated with third-party ordering platforms (and their 30% commissions) built a simple ordering system for his restaurant and two neighboring businesses. Customers scan a QR code, browse the menu, and place orders that appear on a tablet in the kitchen.
What went well: Bolt.new generated a functional ordering flow in a single session. The builder had no idea what React or Supabase were and did not need to. The real-time order updates (powered by Supabase's real-time subscriptions) impressed him most.
What was hard: Payment integration was the biggest challenge. The builder eventually used Stripe Payment Links rather than a full API integration because the AI-generated checkout code had edge cases around failed payments. The app also had no proper error handling for the first month, which caused confusion when the internet went down at the restaurant.
3. PortfolioForge — AI Portfolio Generator
Builder background: UX designer transitioning to product design
Tools used: Cursor with Claude Sonnet
Stack: Next.js, Tailwind CSS, Vercel, Stripe
Timeline: 2 weeks
Status: $800 MRR after 4 months
PortfolioForge lets designers paste their project descriptions and generates a polished portfolio website they can deploy. The builder used her design expertise to create beautiful templates and Cursor to implement the generation logic, template rendering, and payment system.
What went well: The builder's design background was a significant advantage. She knew exactly what the output should look like and could prompt Cursor with specific design references. The AI handled the technical implementation while she focused on the user experience.
What was hard: SEO and performance optimization. The initial version generated portfolio sites that loaded slowly because the AI used large, unoptimized images and did not implement lazy loading. She spent a full week on production optimization after user complaints.
4. TeamSync — Async Standup Tool
Builder background: Engineering manager at a mid-size company
Tools used: Claude Code
Stack: Next.js, Supabase, Vercel
Timeline: 4 days for MVP
Status: Internal tool used by 3 teams (not commercialized)
An engineering manager built an async standup tool for his distributed team. Team members post daily updates via a simple form, and the tool generates weekly summaries and flags blockers. The AI integration uses the Vercel AI SDK to summarize updates.
What went well: Claude Code excelled at building a full-stack application from the terminal. The builder described each feature in sequence, and Claude Code generated the database schema, API routes, and frontend components in a cohesive way. Having programming experience helped him review the generated code and make targeted adjustments.
What was hard: The builder's existing technical knowledge was both a help and a hindrance. He sometimes spent time refactoring AI-generated code to match his preferred patterns instead of accepting "good enough" code that worked. He estimated he could have shipped two days faster by being less particular.
5. GigTracker — Freelancer Income Dashboard
Builder background: Freelance graphic designer, self-taught spreadsheet user
Tools used: Lovable (full build)
Stack: React, Supabase, Vercel
Timeline: 5 days
Status: 200 free users, $0 revenue, abandoned after 3 months
A freelancer built an income tracking dashboard to replace her complex spreadsheet. The app tracked income by client, generated invoices, and visualized monthly earnings. She launched on Product Hunt and attracted 200 signups in the first week.
What went well: The initial build was fast and the product was genuinely useful. Lovable generated a clean dashboard UI with charts and data tables that impressed early users.
What went wrong: This is the honest part. The builder did not implement Row Level Security on her Supabase database, meaning users could technically access each other's financial data by manipulating API calls. A user reported this on the Product Hunt thread. Fixing it required understanding Supabase RLS policies, which took her a week to learn and implement. By then, she had lost momentum and ultimately abandoned the project. The lesson: security basics are not optional, even for MVPs that handle sensitive data.
6. StudyDeck — AI Flashcard App
Builder background: College student studying biology
Tools used: Bolt.new, then Cursor for iteration
Stack: React, Firebase, Vercel
Timeline: 2 weeks
Status: 1,500 users, $500 MRR from premium tier
A student built a flashcard app that uses AI to generate study cards from pasted notes or textbook content. Users paste their study material, the app creates flashcards with spaced repetition scheduling, and a premium tier adds AI-generated practice questions.
What went well: The idea resonated immediately with other students. The builder launched in a university Discord and gained 500 users in the first month without any marketing spend. The AI-generated flashcard quality was the core differentiator.
What was hard: Scaling AI costs. Each flashcard generation call used GPT-4, and at 1,500 active users generating cards daily, the API costs reached $300/month. The builder had to implement caching, switch to cheaper models for simple cards, and add rate limiting — all concepts he learned through trial and (expensive) error.
7. PropMatch — Real Estate Lead Matcher
Builder background: Real estate agent with 12 years of experience
Tools used: Lovable (initial), Cursor (refinement)
Stack: Next.js, Supabase, Vercel, Stripe
Timeline: 6 weeks
Status: $4,200 MRR after 8 months
A real estate agent built a tool that matches property listings with buyer preferences using AI analysis. Agents upload listings, buyers fill out preference forms, and the system generates match scores with explanations. The builder's deep domain expertise was the key differentiator — he knew exactly what matching criteria mattered.
What went well: Domain expertise matters more than technical expertise in vibe coding. The builder's 12 years of real estate experience meant he could write incredibly specific prompts about matching logic that a technical founder would not know. The resulting product felt like it was built by someone who understood the industry because it was.
What was hard: The builder initially tried to build the entire app in Lovable but hit limitations when implementing the complex matching algorithm. He migrated to Cursor after two weeks, which required understanding enough about the codebase to continue development in a different tool. The migration itself took three days.
8. ContentCal — Social Media Scheduler
Builder background: Marketing consultant, former social media manager
Tools used: Cursor with Claude Sonnet
Stack: Next.js, Supabase, Trigger.dev, Vercel, Lemon Squeezy
Timeline: 4 weeks
Status: $1,800 MRR after 5 months
A marketing consultant built a social media scheduling tool specifically for solo consultants (a niche underserved by enterprise tools like Hootsuite). The tool schedules posts, suggests optimal posting times, and generates caption ideas using AI.
What went well: The builder was methodical about prompting. She wrote a detailed product spec before touching any tool, then fed sections of the spec to Cursor one feature at a time. This structured approach produced a more coherent codebase than the "just start building" approach.
What was hard: Social media API integrations. The X/Twitter API, Instagram Graph API, and LinkedIn API all have different authentication flows, rate limits, and data formats. The AI generated integration code that worked in isolation but failed in production due to rate limiting and token expiration edge cases. API integrations remain one of the weakest areas for AI-generated code because the documentation changes frequently and the AI's training data may be outdated.
9. QuoteSnap — Invoice Generator for Contractors
Builder background: HVAC contractor, zero technical experience
Tools used: Bolt.new (full build)
Stack: React, Supabase, Netlify
Timeline: 3 days for MVP
Status: Personal use only, saves 5 hours/week
A contractor built a simple tool to generate professional-looking quotes and invoices from his phone. He enters job details, materials, and labor hours, and the app generates a PDF quote he texts to clients. No subscription, no payment integration — just a personal productivity tool.
What went well: This is the purest vibe coding success story. A non-technical person solved a real problem in his daily work without hiring a developer, learning to code, or paying for a SaaS subscription. The app does one thing and does it well.
What was hard: PDF generation. This is a notoriously tricky area in web development, and the AI-generated code initially produced PDFs with formatting issues on mobile devices. The builder spent two days going back and forth with Bolt.new, refining prompts about PDF layout until the output looked professional.
10. PetBoard — Neighborhood Pet Services Marketplace
Builder background: Stay-at-home parent, former project manager
Tools used: Lovable (initial build), then v0 for UI components
Stack: Next.js, Supabase, Vercel, Stripe Connect
Timeline: 8 weeks
Status: 400 users in one city, $600 MRR from service fees, growing slowly
A neighborhood marketplace connecting pet owners with local dog walkers, pet sitters, and groomers. Providers create profiles, pet owners browse and book, and the platform takes a 10% service fee. The builder used Lovable for the initial marketplace structure and v0 to generate polished UI components.
What went well: The two-sided marketplace structure — provider profiles, search/filter, booking flow, payment splitting — would have cost $15,000-$30,000 to build with a freelance developer. The builder spent approximately $200 on tool subscriptions over two months.
What was hard: Stripe Connect (the platform payments product that splits payments between marketplace and providers) was the single hardest integration. The AI generated code that worked for simple payments but did not properly handle the connected account onboarding flow, payout schedules, or dispute management. The builder hired a freelance developer for 10 hours ($500) to fix the Stripe Connect integration specifically. This is a common pattern: vibe code 90% of the app, hire an expert for the remaining 10% that requires deep domain knowledge.
Patterns Across All 10 Projects
Several patterns emerge from these profiles:
- Domain expertise is the differentiator. The most successful projects were built by people who deeply understood the problem they were solving. The AI handled the code; the builder provided the insight.
- The 90/10 rule. Vibe coding tools handle roughly 90% of most projects well. The remaining 10% — payment integrations, security, performance optimization, third-party APIs — requires either learning or hiring help.
- Security is a real risk. Multiple projects had security issues that the AI did not flag proactively. Security basics should be part of every vibe coder's checklist.
- Simple beats complex. The most satisfying projects (QuoteSnap, LocalMenu) solved specific problems simply. The ambitious projects (PetBoard, ContentCal) took significantly longer and required more technical depth.
- Tool migration is common. Several builders started with an AI app builder (Lovable, Bolt.new) for speed and migrated to an AI code editor (Cursor, Claude Code) when they needed more control. This is a healthy, common pattern.
- Revenue is achievable. Five of the ten projects generate revenue. The amounts are modest ($500-$4,200 MRR), but they represent real businesses built by non-developers with minimal upfront investment.
What You Can Build Next
If these examples inspire you, here is how to start. Pick a problem you personally experience or deeply understand. Do not build a generic "todo app" — build a tool for your specific workflow, industry, or community. Your domain expertise is the one thing AI cannot replicate, and it is the ingredient that turns a side project into a product people will pay for.
Start with choosing your first tool, describe your idea clearly, and build your first app. If these ten builders can do it, so can you.