Update your writing profile

The Writing Profile is the document that captures how you write: every draft in your workspace starts from it, and every draft is checked against it. It learns from the feedback you give on real articles, and until auto-learn launches (soon), you trigger that learning yourself with one button. This guide covers why the profile matters, how to update it, and what the rest of Writing Calibration controls.

Why the profile matters

Every time you review an article, you produce two kinds of signal: the annotations you leave on the draft and the edits you make to it directly. The same goes for anyone else who reviews in your workspace, whether they work inside the editor or through a share link. Together, that markup is the most honest record of your voice the workspace ever gets, because it shows exactly where the drafts and your taste diverge.

When you update the profile, it reads that signal (the annotations and changes on your reviewed articles) along with your workspace context, and rebuilds itself around what you actually corrected. That is what makes the profile compound: a correction applied to one draft fixes one page, but a correction absorbed into the profile fixes every page that comes after. A profile that has absorbed a few review cycles writes drafts that need less review; a stale one keeps making the same mistakes, and you keep fixing them by hand.

How to update the profile

If several people on your team review drafts, it helps for everyone to know the loop exists: your edits and annotations teach the system, and updating the profile is how the lesson sticks. People who know their markup gets absorbed leave better markup.

When to update

• After the first review cycle, always. The first batch of feedback is the richest correction the profile will ever get, because it is the first contact between the starting profile and your real taste.
• After deep edits to an article. When you spend real time editing and annotating a piece with important feedback, that is exactly the signal the profile should absorb. Update so it carries forward.
• Whenever a batch of articles clears review with real edits or markup. Real corrections mean real divergence; fold them in before the next batch repeats them.
• When you spot repeat feedback. Making the same correction on two drafts is the signal you waited one update too long.
• Not after every article. Feedback on a single draft can be noise, and one-off preferences do not deserve a permanent place in the profile. Let signal accumulate, then update.

When auto-learn ships, this trigger becomes automatic and the profile keeps learning on its own.

The rest of Writing Calibration

The profile is what the workspace writes from; Writing Calibration is the set of inputs you tune, and sharper inputs give every update a better starting point. It sits alongside the profile in Context, and covers:

• Reference writers. The writers whose style sets your bar. If you keep rewriting openings to sound like someone you admire, add that writer here.
• Tone. How the writing should carry itself: the stance, the formality, the energy. If tone is what you keep fixing, this is the dial.
• Source hierarchy. The tiered list of sources drafts should lean on, from your own material outward. It controls what the writing cites and trusts.
• Example pieces. Work that already sounds like you. If you shipped a piece you love since the last calibration, add it so the profile anchors on it.
• How you describe what you sell. The language the writing uses for your product and category, in your words rather than generic ones.

You tune calibration; the workspace writes from the profile. The full picture of where calibration sits in Context is in the calibration guide.

Signals it worked

Your feedback on voice and tone drops off over the following drafts, and what remains shifts toward facts and specifics, which is where your review time actually earns its keep. You stop re-litigating the same corrections. And when auto-learn ships, your workspace inherits a profile that has already been learning, not one that needs a rescue.

Where to go next