The fastest way to burn AI credits is not a long chat. It is rebuilding something you already shipped because the brief was thin, the path was unclear, or someone wanted a “cleaner” version for transport.

If the outcome already exists as files, a page, a webhook, or a prompt pack, your next move is reuse. Rebuild is the expensive exception, not the default.

The tip

Before you open a builder, a cloud agent, or a fresh project, answer three questions in writing:

  1. What already exists that gets us most of the way?
  2. What is actually broken or missing (not “I wish it lived somewhere else”)?
  3. Can we re-point, upload, or patch instead of regenerating?

If you cannot name the reuse path, you are not ready to spend a build turn.

What counts as reuse

  • Re-pointing a host form or webhook at a working endpoint
  • Uploading files that already pass review
  • Copying a known-good prompt or template into a new desk
  • Patching one function or one page instead of scaffolding a twin app
  • Linking an existing draft instead of rewriting it from memory

Transport problems (where the files live, how they get to the host) are not a license to regenerate the product. Move the artifact. Do not reinvent it.

When rebuild is honest

  • The live version is wrong in substance, not just location
  • Security or credentials force a clean cut
  • You explicitly decided the old approach is retired

Even then, the brief should list what not to redo: paths to keep, copy to preserve, and the success check that proves the new thing is better than the old one.

Paste-ready rule

Reuse-before-rebuild rule:
1) Before any build, list existing paths/files/pages that already solve the outcome.
2) Prefer re-point, upload, or patch over greenfield.
3) “Needs a new home” is a deploy task, not a rebuild.
4) Only rebuild when substance is wrong, security requires it, or the old approach is retired on purpose.
5) Every build brief must include: outcome, reuse list, do-not-rebuild list, success check.

This keeps the desk useful for judgment and drafting, and keeps the expensive build turns pointed at new value instead of remaking last week’s work.

Reuse Before You Rebuild