February 9, 2026
When prompts meet office politics
Why "just prompt better" doesn't work
‘Just Prompt Better’ Backfires: Devs Blame Meetings, Misreads, and AI Overconfidence
TLDR: The piece argues AI code tools save typing time but create more review and rework because real issues are human and contextual, not just code. Commenters split between “make assistants ask smarter questions” and “the real fix is better communication with stakeholders,” making this a people problem, not a prompt problem.
Developers are roasting the idea that “just prompt better” will fix AI coding tools. The article says what many devs feel: AI can spit out code fast, but the real chaos lives in people problems—missed context, vague plans, and decisions made in meetings that never make it back to the code. One-third of tech “gotchas” pop up during planning, and 70% need to be explained to non-coders, turning Slack threads into soap operas. Half the gotchas only show up once someone’s actually building the thing—aka the “devil in the details” phase that AI tends to bulldoze through.
That’s where the drama explodes. One commenter says the “central tension” is that code generators speed past the very step where devs usually discover hidden constraints. Another blames the chaos on stakeholders without a clear picture of what they want, turning every feature into a game of telephone. Optimists argue assistants can ask questions and even refuse bad solutions; skeptics clap back that they’ve tried fancy planning add-ons and the counterproposals still aren’t “staff-level.” The memes write themselves: “Just prompt harder,” “AI as the intern who never asks the right questions,” and “bring your assistant to the standup.” Whether you’re Team “Smarter Questions” or Team “It’s the Meetings,” the vibe is clear: this isn’t a typing speed problem—it’s a context problem. source
Key Points
- •The follow-up article analyzes 40+ developer survey responses about AI coding assistants’ impact on software development.
- •Atlassian (2025) is cited: AI adoption increases time spent on review, rework, and realignment, offsetting implementation time saved.
- •About one-third of technical constraints are discovered during product meetings and planning sessions.
- •Approximately 70% of constraints must be communicated to stakeholders who do not regularly interact with the codebase.
- •Around half of constraints are discovered during implementation, which functions as context discovery; downstream realignment is costly due to communication latency and redundant conversations.