The article highlights key issues developers face with coding assistants, particularly how reliance on AI can exacerbate communication problems in software development. A significant finding is that one-third of technical constraints are discovered during planning sessions, but communicating these constraints to stakeholders is often cumbersome. A large portion of developers noted that 70% of these constraints need to be communicated to people outside the codebase, leading to unresolved issues until they become critical during implementation. The article emphasizes the difficulty in articulating complex dependencies and translating technical issues into business impacts, which ultimately slows down projects. Another critical point raised is that while AI can generate code, it fails to challenge poor decisions or suggest better alternatives. Developers reported that AI lacks the context and authority to push back on prompts, limiting its effectiveness in collaborative environments. The article argues that simply encouraging developers to "prompt better" doesn't address the underlying issue that many constraints are only identified during coding. This highlights the need for human insight during the planning phase, as AI tools tend to follow instructions too literally without considering broader implications. Overall, the article calls for better communication strategies and tools that facilitate cross-functional alignment rather than just focusing on code generation.