In a company presentation recently, one of our PMs said, I can't vibe code, so...
— and I haven't stopped thinking about it since.
The phrase was meant as a lighthearted self-own, but it hit me like a revelation: PMs have been vibe coding all along. They write feature requirements in natural language, share them with implementation agents (aka developers), and review the output to see if it matches their intent — often without ever diving into the underlying technical details. Sound familiar? It should. It's exactly how prompt engineering works with AI.
Consider the typical PM workflow: drafting detailed requirements, creating user stories, building wireframes, and articulating acceptance criteria. These are essentially sophisticated prompts designed to guide developers toward a desired outcome. The PM doesn't specify how to implement the feature any more than a prompt engineer tells the LLM which neurons to fire. Instead, both are focused on describing what the end result should accomplish and how it should feel.
This parallel becomes even more striking when we examine how PMs handle feedback cycles. When a feature doesn't quite match the vision, a PM doesn't rewrite the code — they clarify the requirements, highlight the misalignment, and ask for adjustments. This iterative refinement process mirrors how we work with AI models today: prompt, evaluate, refine, repeat. In both scenarios, the focus is on steering outputs through better inputs rather than direct manipulation of the underlying system.
What we're really talking about is a fundamental skill that transcends both product management and AI interaction: the ability to express intent clearly enough that another entity can transform it into reality. This is the essence of vibe coding
— communicating a vision with sufficient clarity and context that it can be implemented by someone (or something) else. It's less about syntax and more about conveying the right feel, the right user experience, the right outcome.
As AI continues reshaping our workflows, this skill becomes increasingly valuable across all roles. PMs may find themselves uniquely positioned for this future, having honed the art of translating human needs into implementable specifications for years. Meanwhile, developers might evolve toward becoming interpreters of increasingly high-level intent, adding their expertise to refine and implement the vision rather than translating explicit instructions into code.
The rise of vibe coding doesn't diminish the importance of traditional coding — complex systems will always need experts who understand their inner workings. But it does raise fascinating questions: If expressing intent becomes more powerful than writing implementation, how does that reshape our organizational structures? If everyone can vibe code through AI interfaces, where does the PM role end and the developer role begin? Perhaps the lines between these disciplines are blurrier than we thought, connected by a shared ability to translate human needs into functional systems.
So maybe PMs don't need to learn to code
in the traditional sense. Maybe they've been doing their own kind of coding this whole time — vibe coding — and that skill might just be the most future-proof one of all.