Don’t draft on moving ground.
Most document rework isn’t bad writing. It’s good writing done against a brief, a price or a scope that changed underneath it.
Watch a proposal come together in most organisations and you will see the same pattern. Drafting starts early, because the deadline is close. The scope is still being argued. The pricing is “nearly final.” Two people are pulling evidence from different folders. By the time the document reaches review, half of it describes a job that no longer exists.
The reviewer’s comments are not really about the writing. They are about the inputs: this rate changed, that client requirement was dropped, the case study refers to a project we decided not to mention. Every comment sends a writer back to rework sections that were fine when they were written.
AI speeds up the wrong part
AI drafting makes this worse before it makes it better. When a section takes a minute to generate, it is tempting to generate it immediately and fix it later. But “later” is where the cost lives. A fast draft built on unsettled inputs is just a faster way to produce rework, and a harder one to review, because nobody can tell which version of the facts each paragraph assumed.
Settle, then write
The fix is old-fashioned discipline, made practical. Settle the inputs in order, and lock each one before the next begins.
The brief. What is being asked for, by whom, by when, and what a good answer must cover. Agreed once, then fixed.
The evidence. The specific documents, figures and past work the response will draw on. Chosen deliberately and sealed, so the drafting can’t wander into material nobody approved.
The plan. The sections, what each must say and roughly how long it should be. Locked before a word of prose is written.
Only then does drafting start, section by section, each against the same fixed foundation. If something genuinely changes, you don’t patch paragraphs: you reopen the input that changed and redraft the sections that depend on it, on purpose.
Speed comes from settling the ground first. Writing faster on moving ground just moves the rework downstream.
Teams that work this way find something counter-intuitive: drafting starts a little later but the document finishes sooner. Review is shorter, because reviewers are checking judgement rather than chasing facts. And the finished document comes with its foundations attached, which is exactly what you need when someone later asks what was sent, and why.