Steipete reports that with GPT-5.5, /goal, autoreview and crabbox, his prompts moved from 30-60 minute tasks to often 4-10 hour tasks, and his confidence that the result is ready is much higher. His one-liner: 'Yielding agents is a skill.'

With 2905 likes, this resonates deeply—the shift from micro-managing agents to trusting them with long-running tasks is the emerging meta-skill of the AI era.

Steipete shares a practical trick with Codex: ask it to review code for bugs and it will say all good. Tell it there IS a bug and it will loop and loop and find issues.

This is a subtle but powerful prompt engineering insight: framing the same task differently—neutral review vs. assuming a bug exists—produces dramatically different results. 2698 likes.

Dan Shipper shares his Codex usage stats: 38 billion tokens processed, longest task ran for 56 hours, 41-day current streak. His sign-off: 'lfg codex'.

These numbers are remarkable—56 hours of autonomous agent work means the agent is handling entire project-level tasks, not just snippets.

A user highlights what they love about auto-review: Cursor explains the command and risk, making it much easier for new coders to learn and just do things.

This is the educational layer of agentic coding—the agent doesn't just do the work, it teaches you why it's doing what it's doing.

Thibault Sottiaux (OpenAI) explains the versioning strategy: going from GPT-5.0 to 5.1 to 5.5, the incrementing goes with improvements in capabilities and token efficiency (which translates to speed gains). GPT-5.5 is their best model yet. A simple strategy they want to continue.

With 3842 likes, this is the most-liked tweet of the day. The framing is clear: version numbers map to real capability and efficiency improvements.

Guillermo Rauch's 2599-like one-liner: Ship the best product. Use lots of AI, some AI, maybe no AI. Just be the best.

This is a corrective to the AI-everything hype: the goal is not to use AI for its own sake, but to make the best product possible. AI is a tool, not the end.

Aaron Levie continues his jobs crusade with enterprise field data: in the majority of conversations with CIOs, CTOs, and CEOs in large enterprises, they are either growing due to AI (new job functions like FDEs, engineering) or reinvesting efficiency savings back into the business.

He cites Goldman Sachs CEO David Solomon's NYT OpEd: the AI boom creates new jobs in the build-out and implementation, while freeing up dollars to invest in areas that have been underfunded. Most businesses have been constrained by how much software they can produce, how many sales reps they can hire, how many campaigns they can run. When AI makes it possible to do more, investment goes back into the business.