OpenAI just turned its elegant Mac app into a 1.5 GB Electron monument to feature bloat while simultaneously hiring someone to make ChatGPT safe for grandma.

The Summary

The Signal

OpenAI's product announcement created a desktop app situation so confusing that a tech journalist literally quoted Airplane while trying to explain it. The old ChatGPT Mac app, a 159 MB native application, is now "ChatGPT Classic." The new ChatGPT desktop app is actually Codex, but also ChatGPT Work, and also includes pieces of the dead Atlas browser, all wrapped in 1.5 GB of Electron wrapper. Users can now install ChatGPT Classic, ChatGPT, and Codex simultaneously, though the "way forward" is unclear.

This is what happens when you try to be everything to everyone at once. The technical debt is visible in the file size alone. A tenfold increase from 159 MB to 1.5 GB doesn't suggest feature richness. It suggests architectural compromise.

"The ChatGPT Classic app looks more native Mac-like, so that might be an issue for users."

Meanwhile, OpenAI is hiring for families. Not just users, not just enterprises. Families. The job posting asks for someone to build experiences for caregivers and older adults. This is the consumer play, the household penetration strategy, the "ChatGPT in every kitchen" vision. It's also fundamentally at odds with shipping a 1.5 GB power-user Frankenstein app on the same day.

The family product manager will be building for people who don't know what Electron is, don't care about technical modes versus abstracted modes, and definitely don't want to choose between three differently named apps that do similar things. They want an app that works, loads fast, and doesn't make them feel stupid. The new ChatGPT desktop app is none of those things.

Key tensions emerging:

  • Native Mac experience (fast, light, familiar) versus cross-platform Electron (slow, heavy, generic)
  • Technical users who want Codex depth versus families who want simple, safe AI assistance
  • Product velocity (ship everything now) versus user experience coherence (make one thing great)

The Atlas browser absorption is particularly telling. Atlas was OpenAI's attempt at an AI-native browser, discontinued after failing to gain traction. Rather than cleanly retiring it, they've folded its code into the ChatGPT superapp, adding weight and complexity for features most users will never touch. This is how products become bloated. Not through malice, but through the accumulation of strategic pivots that never get properly pruned.

The Implication

Watch what OpenAI does next with ChatGPT Classic. If they deprecate it within six months, the native Mac experience dies and Electron wins by default. If they maintain both apps in parallel, that signals they know they screwed up and are buying time to fix it. The family hire suggests they understand the consumer opportunity, but consumer products don't win by being heavier and more complex. They win by being simpler than what came before.

For anyone building AI products: this is a case study in what not to do. Consolidation for its own sake creates complexity for users. Feature addition without subtraction creates bloat. Technical users and mainstream users have different needs. Trying to serve both in one app usually means serving neither well.

Sources

TechCrunch AI | Daring Fireball