Codex is strong, but the product line still feels behind Claude in agent UX, integration, and pace

What variant of Codex are you using?

App,CLI

What feature would you like to see?

I’m writing this as a frustrated-but-still-hopeful Codex user.

The frustration is exactly because Codex is not weak. The foundation is clearly strong. That’s what makes the current gap more painful.

From a user point of view, OpenAI should be in the strongest position here: stronger model base, bigger company, broader product surface, more engineering resources. So why does the Codex product line still feel slower, less integrated, and less ambitious than Claude’s product line?

That is the honest user impression right now.

Codex is good at the core coding workflow. But Claude increasingly feels like an agent system that is escaping the “coding assistant” box:

• remote control
• browser-side workflows
• computer-use framing
• stronger continuity across surfaces
• a more visible extension/product ecosystem
• a bigger sense of product direction

Codex, meanwhile, often feels like a collection of strong components that have not yet fully snapped together into a single compelling agent experience.

That is the part I hope the team takes seriously.

This is not a complaint that “Codex has no features”.
It absolutely has features.
The issue is that competitors are increasingly shipping integrated user experiences, while Codex still often feels like:

• app here
• CLI there
• other agent capabilities elsewhere

As a user, I don’t judge by org charts or internal architecture. I judge by the felt experience:
“Does this product already feel like the future?”

Today, Claude often feels closer to that than Codex does.

My biggest requests:

1. Make Codex feel like one continuous agent across app / CLI , not several partially connected surfaces.
2. Push harder on remote-control and session continuity.
3. Bring computer-use / browser-action workflows into the foreground in a Codex-native way.
4. Invest more aggressively in memory, continuity, and repeated-use adaptation.
5. Build a stronger extension ecosystem around hooks/plugins/discovery/sharing.
6. Ship with a clearer sense of product convergence and momentum.

I’m saying this because I want Codex to become the obvious best product in this category.

Right now, it has the ingredients, but not yet the product synthesis.
And from a user perspective, that makes OpenAI feel like it is reacting to the market more than defining it.

I hope that changes.

Additional information

_No response_