May 18, 2026
Your coworker is now a chatbot
Codex-Maxxing
AI work buddy or office nightmare? Readers are wildly split on Codex-Maxxing
TLDR: OpenAI’s Codex is being pitched as more than a coding tool: a persistent assistant for notes, projects, and office tasks. Readers were split between calling it a breakthrough for messy real-world work and warning it feels creepy, unreliable, and a little too much like replacing your brain with office automation.
One writer says OpenAI’s upgraded Codex app has evolved from a code helper into a kind of all-day digital work sidekick: it remembers long-running projects, takes messy voice notes, keeps working while you keep talking, and even helps draft replies for Slack and email. In plain English, the dream is simple: dump your chaotic brain into the app, let it organize your work, and come back to something useful.
But the comments? Absolutely not calm. One camp was instantly sold on voice mode, with people basically yelling, “Why is nobody using this more?” The idea of rambling your half-formed thoughts out loud and letting the system turn them into plans was seen as a huge unlock. Another crowd went full productivity-goblin, sharing elaborate setups with multiple named agents and “manager” bots coordinating work like a tiny AI office.
Then came the backlash. The sharpest criticism was that these systems can look organized while quietly making stuff up, skipping updates, or replacing real notes with flimsy filler. One commenter warned that without careful checking, your neat little memory vault can drift away from reality. And the biggest drama bomb landed when someone called the whole Slack-and-email drafting loop “dystopian,” basically saying: please never make me work with a person whose robot is pre-reading and pre-writing office chat. Add in the not-so-small detail that the author works on the Codex team, and readers were suddenly side-eyeing the whole post like it was productivity advice and an ad.
Key Points
- •The article describes using Codex not only for software development tasks but also for broader knowledge work such as presentations, notes, documents, and other artifacts.
- •A key workflow change is the use of durable long-running threads supported by compaction, allowing ongoing workstreams to retain history, preferences, and prior decisions.
- •The author says long-running threads can cost more when old context is no longer cached, but considers the continuity worthwhile for important tasks.
- •Voice input and transcripts are presented as useful because they capture unedited thinking and messy source material that can improve planning and writing.
- •The article argues that effective long-running AI workflows need shared memory outside a single thread or repository, with knowledge serialized into durable, inspectable artifacts such as files in an Obsidian vault.