Transparency Center

What we publish about how we operate.

Transparency is the part of trust we can control. This page collects the decisions, practices, and disclosures we believe a serious reader has a right to see — and the things we are still working out in public.

The models we use

The rewrite engine routes between a small set of frontier language models depending on the task. We do not train our own foundation models. We do tune prompts, build evaluation harnesses, and maintain a layer of editorial guardrails on top of the underlying models. When we change a default model — for cost, latency, or quality reasons — we note it in our changelog rather than silently swapping.

What we do with your content

Your drafts, rewrites, and stored documents are yours. We do not use your content to train any model — ours or any third-party provider's. We do not sell access to your content. We retain content for the time needed to provide the product (and the time you choose to keep it stored), and we delete on request. Specific data handling details are in our privacy policy.

The team and the company

RewriteAIForMe is built by a small editorial-and-engineering team. Our editorial work is led by working writers and editors; our engineering work is led by people who have shipped production-grade language-model products. Bios, named authors, and the company entity behind the product are on the about page. We update both when they change.

Editorial independence and incentives

We do not accept paid placement in editorial content. Articles in the Academy, Guides, and Research sections are written by named contributors against the same editorial standards we apply to ourselves. When a piece references a product — ours or another — we say plainly that we make a tool in this category and disclose any commercial relationship that exists.

Corrections and material changes

Errors in published content get corrected with a visible note. Material changes to product behaviour, pricing, data practices, or terms are announced ahead of taking effect, with enough notice for users to act on the change. The pattern we are building toward is closer to a public service's communication discipline than a SaaS company's release-notes style.

What we are still working out

Some questions do not yet have a clean answer — most notably the long-term ethics of training-data attribution, the right disclosure standards for AI-assisted writing across genres, and the best way to surface model uncertainty inside the product. We will write about each as we develop a position we can defend. We would rather publish the working answer than the polished one.