What Can You Actually Build? 10 Real-World Examples

Everyone's talking about vibe coding. Almost nobody is being specific about what that means in practice. So let's fix that.

Over the past year, we've tracked what people are actually building — not demos, not proof-of-concepts, but real products with real users. Here are 10 of them, with honest observations about what worked and what didn't.

1. SaaS Dashboards

This is where vibe coding shines brightest. Analytics interfaces, admin panels, reporting tools — anything that's essentially a data model connected to a visual layer. A solo founder in Stockholm built a competitor analysis dashboard for e-commerce brands and reached 200 paying customers in three months. Her background was in marketing, not engineering. She'd never written a React component before.

2. Internal Business Tools

Companies are quietly replacing their spreadsheet workflows with proper software. One logistics firm we spoke to replaced a 12-tab Excel monstrosity with a custom route optimization tool in about two weeks. It's not glamorous, but it saves them four hours a day. This is where some of the most underappreciated value is hiding.

3. AI Chatbots with Custom Knowledge Bases

Retrieval-augmented generation (RAG) chatbots — tools that answer questions based on your specific documents — have become genuinely accessible. A Norwegian law firm built a tool that answers client questions using their internal documentation. Not perfect, but it handles 60% of routine queries automatically. The implementation, including chunking, embeddings, and a simple UI, took five days.

4. E-Commerce Sites with Unusual Requirements

Generic Shopify works fine for generic stores. When requirements get specific — custom pricing logic, unusual inventory structures, bespoke configurators — vibe coding closes the gap. A manufacturer building custom furniture needed a 3D configurator with real-time pricing. That would have been a 40,000 EUR project two years ago. They built it themselves in three weeks.

5. Data Visualization Tools

Turning raw data into interactive charts and insights is still hard enough that most teams just use screenshots from Excel. Observable and similar tools changed this somewhat, but the real shift is in building custom viz tools for specific domains. We've seen financial analysts, climate researchers, and sports teams build bespoke visualization layers on their own data with no engineering support.

6. Booking and Scheduling Systems

Calendly clones with extra requirements, multi-location scheduling for service businesses, complex resource allocation tools — these are built in days now. One physiotherapy clinic built a full booking system with automated reminders, insurance tracking, and a patient portal. Acuity Scheduling would have cost them 500 EUR/month. The custom build cost three days of work.

7. Mobile App Prototypes

Full native apps remain genuinely hard to build solo. But React Native prototypes that look and feel nearly production-ready? That's now accessible. Several founders have used vibe coding to build prototypes compelling enough to raise seed funding — then hired engineers to productionize them. The prototype-to-funding pipeline has shortened considerably.

8. Automated Reporting Pipelines

Connecting data sources, transforming them, and producing formatted reports on a schedule — this kind of workflow automation was always possible but tedious to build. Now it's a weekend project. Marketing agencies are building client-specific reporting tools instead of manually compiling numbers every month.

9. Developer Tools and CLIs

This one surprises people. Developers using AI assistance to build tooling for other developers. Custom linters, code generators, deployment scripts, testing frameworks. Agentic workflows where one AI model orchestrates others are particularly interesting here — the toolchain is becoming recursive.

10. Content and Media Tools

Transcription pipelines, content moderation systems, image tagging workflows, SEO analysis tools. A podcasting network built their entire post-production pipeline — transcription, chapter generation, show notes, social clips — in about two weeks. It replaced 20 hours of manual work per episode.

The Pattern

Looking across all of these: the common thread isn't the category — it's the specificity. The tools that get built are specific to particular problems in particular contexts. Not generic solutions that compete with established SaaS, but deeply contextual tools that solve a specific pain precisely. That's what becomes economically viable when building is cheap.

What specific, niche problem in your world has always been too small to justify proper software? That's where the interesting builds are happening.