ZDNET tester builds a basic text-analysis app with Lovable after trials of Cursor and Replit

A ZDNET writer experimented with AI "vibe coding" tools — including Cursor, Replit, Microsoft Visual Studio with GitHub Copilot, and Lovable — to try to build a simple data-analysis app on a Mac, and ultimately produced a minimal viable product using Lovable. The reviewer found these tools can automate many setup tasks such as creating project files, setting up virtual environments, installing libraries and launching local servers from a natural-language prompt.

But problems emerged: Cursor restarted and lost chat history stored in a workspace on the Mac and offered no way to force retrieval, which led the writer to abandon it. Replit, a cloud-only service, set up an environment quickly but the reviewer hit the free-plan quota before meaningful work could be completed and encountered errors extracting content from Apple .Pages files.

Visual Studio on the local machine produced only a basic text-string–matching function despite hours of prompting Copilot, not the deeper textual analysis the writer wanted. Lovable provided the fastest front end and the simplest interface; pairing it with Google Gemini and switching data to an XML archive on Algolia allowed the reviewer to obtain rudimentary semantic analysis and arrive at a working prototype after upgrading to Lovable's paid plan.

The experiment highlighted concerns about cloud-hosted tools charging credits quickly and the need to expose data to remote servers.

Key Topics

Tech, Lovable Labs, Cursor, Replit, Visual Studio, Github Copilot