I’ve been using Cursor obsessively lately, which sparked some ideas so I want to share them here

Conclusion first: with AI-assisted capability, one person becomes an entire team

The legacy development routine

PRD + prototype + code

  • Product managers interview stakeholders, nail down requirements, and produce the PRD
  • Designers coordinate with product to deliver the UI prototype
  • Engineers align with product and design, clarify requirements, then schedule and start coding

This is the typical mainstream flow today (excluding tiny workshops). The whole process involves many roles and inevitably produces plenty of meetings, back-and-forth discussions, and alignment sessions—you know the drill.

However, after diving into Cursor and Claude recently for work, I suddenly realized that this whole model might disappear at lightning speed.

Demo of development

Take a “management platform” project for example; the page looks like:

Picture demonstration
  • At the interaction level the interface is clean, silky-smooth
  • At the code level every module matches the requirements doc precisely. The code itself looks very professional—functions are carefully packaged and the comments are thorough

Final takeaways

It’s not perfect, but a 90 out of 100 is still worth celebrating. That project—from PRD clarification to final release—only took half a day. And a chunk of that was wasted fighting network issues.

By leveraging AI, production speed skyrockets. There is a trending notion overseas called “one-person company,” and I believe that will become the future development model. The entire team might literally be one person, orchestrating everything through agents (frontend, backend, UI, DevOps, QA…). The person at the center of decision-making needs to focus on direction, requirements, and goals; the specific technical implementations become irrelevant. Of course, the practice still has its pain points—loop issues, mismatched descriptions vs. actual changes, lost-in-translation situations—and current AI tools may not yet reach our ideal level of perfection. But I believe that’s only a matter of time, because AI capability keeps evolving.

For frontline developers, this is both a challenge and an enormous opportunity. We need to shift our self-perception and gradually become “decision-makers.” Since today’s AI still cannot fully support “natural language input → finished output,” experienced developers who understand the entire delivery chain instantly become perfect conductors. Mastering AI will be our future role.

The era of old-school programming is ending. The author of The Mythical Man-Month would never have expected that years later, AI—nearly omnipotent—would deliver a true “silver bullet,” and we are entering a new generation at breakneck speed.

Picture description