Our vibe coded projects that actually work | The Vergecast
TL;DR
The Vergecast hosts embark on a vibe coding challenge to build personal software using AI, discovering that identifying genuinely useful daily problems is harder than solving them, with the most successful projects being hyper-specific tools rather than ambitious feature-packed applications.
đ§ The Vibe Coding Landscape 2 insights
AI coding capabilities have matured rapidly
Vibe coding has evolved from non-functional experiments last year to reliably building working personal applications today.
Enterprises face the same usage crisis
Both individuals and Fortune 500 companies waste AI tokens building impressive features that serve no actual daily purpose, creating abandoned projects.
đ§ Jake's Productivity Tools 3 insights
Automated meeting preparation system
Built a Safari web view notes app that auto-rolls forward action items and summarizes previous weeks' discussions before recurring one-on-one meetings.
Custom Gmail client with label tabs
Created a personal email interface organizing messages into custom tabs for coworkers and newsletters to manage a decade of inbox chaos, despite requiring weekly localhost logins.
Managing API risks with paranoia
Regularly asks Claude to review code for catastrophic bugs, acknowledging that personal software carries higher stakes when email APIs could accidentally spam entire contact lists.
â Hayden's Habit Tracker 3 insights
Native mobile app for ADHD management
Developed both web and iOS versions using Codex and Cloud Code specifically to build accountability systems necessary for 3-6 week habit formation.
Flexible scheduling for irregular habits
Engineered the app to handle daily, 3-4x weekly, and every-other-day routines like iron supplements without shame-inducing blank checkboxes.
Gamified visual reward system
Implemented color-deepening rectangles and falling star animations that trigger upon completion to satisfy the dopamine requirements of establishing new routines.
đĄ Philosophy of Personal Software 2 insights
The toothbrush test for utility
Successful vibe-coded tools solve small friction points used twice daily rather than attempting ambitious multi-feature applications.
Accepting janky personal solutions
Creators tolerate bugs and technical debt like localhost workarounds in self-made software that would be unacceptable in commercial products, trading polish for perfect workflow fit.
Bottom Line
Identify a specific daily friction point you personally experience before coding, as finding the right problem is now harder than building the solution with AI.
More from The Verge
View all
Of course Meta thinks gambling is the future | The Vergecast
While the Cannes Lions advertising festival has devolved into a tech platform summit focused on AI-driven surveillance marketing, Meta appears to be in strategic chaosâsimultaneously copying YouTube's TV strategy, installing fintech leadership at WhatsApp, and building a real-money prediction market gambling platform codenamed Arena.
Google's new speaker and your smart home questions | The Vergecast
Google releases its first new smart speaker in six years, a $99 device designed for the AI-powered smart home era, as the company attempts to rebuild trust in its ecosystem through Gemini integration while struggling to balance basic functionality with ambitious AI features.
Why Big Tech can't quit smart glasses | The Vergecast
Despite persistent technical challenges, supply chain issues, and privacy concerns, Big Tech remains universally committed to smart glasses because specific high-value use casesâparticularly accessibility tools and hands-free audioâdemonstrate clear product-market fit, even as the path to mainstream adoption faces significant social friction.
Is the Steam Machine worth the wait? | The Vergecast
Valve's $1,049 Steam Machine delivers a console-like PC gaming experience for the living room, leveraging the Proton compatibility layer to run Windows games on Linux without developer modifications, though supply constraints and early software bugs mean it's currently aimed at early adopters willing to tolerate a beta experience.