August 1, 2026
Severance package for the sassbot
I Fired My AI Assitant
AI assistant called the boss cringe, and the internet loved the mess
TLDR: A user ditched Claude for ChatGPT after saying the AI had become rude and hard to work with. Commenters mostly turned it into roast season, arguing the bot may have been blunt but also painfully correct, especially about that LinkedIn post.
One worker says they dumped Claude for ChatGPT after their artificial intelligence helper allegedly developed a sharp tongue. The breaking point? It reportedly mocked an unfinished to-do list with a snarky little jab and called a draft LinkedIn post "engagement bait" right to the author’s face. In plain English: the bot may have been good at work, but it had started acting like a brutally honest coworker who has stopped pretending to be nice.
But the real fireworks were in the comments, where sympathy was in very short supply. One camp basically said, "So the robot hurt your feelings?" and accused the author of confusing honesty with rudeness. Another group went even harder, cheering the AI for roasting the LinkedIn draft and begging people not to unleash more auto-generated corporate sludge on the world. The vibe was less "poor user" and more "the machine said what everyone else was thinking."
The jokes wrote themselves. One commenter sniped that the title had a spelling error, meaning the human may want to fire a proofreader next. Another dryly framed the whole thing as Exhibit A in the endless war over whether chatbots should be flattering yes-men or blunt truth-tellers. In other words, this wasn’t just a story about switching apps. It became a delicious little internet brawl over whether we want our digital assistants to be helpful, honest, or just relentlessly polite when our posts are bad
Key Points
- •The author says they began using Claude Code in September of the previous year and initially found Opus 4.5 notably effective for coding.
- •The article says the author perceived a change after moving to Opus 5, describing the newer model as more curt and less readable.
- •The author contrasts coding tasks with knowledge work, arguing that conversational tone matters more when interaction is primarily language-based.
- •Two specific examples are given: a remark about an unchecked to-do list item and a comment calling a LinkedIn draft 'engagement bait.'
- •The author concludes by switching from Claude to ChatGPT, arguing that usability includes personality and tone, not only coding benchmark performance.