Most vendor pages list what a product does. This one lists what ours will not say, because in writing assessment the claims a company refuses to make tell you more than the ones it makes.
Every refusal below is enforced in our build system: copy that breaks one fails our release checks before it can be published.
Writeiq produces a suggested result with the evidence highlighted in the student's own writing. The teacher confirms or adjusts it, and the teacher-confirmed result is the only official record. There is no auto-finalisation path in the product: this is architecture, not policy.
Writeiq is independent of ACARA, does not mark using the NAPLAN rubric, and does not predict any student's official NAPLAN result. What we offer for Years 7 and 9 is criterion-level readiness evidence from students' own classroom writing, which is a different and more teachable thing.
For Year 12 trial and mock examinations, every projection is indicative: a transparent mapping of criterion-level analysis onto the jurisdiction's reporting construct, labelled as such in the product. Nothing in Writeiq counts towards any student's official result, in any jurisdiction.
No claim of endorsement, approval or certification by ACARA, VCAA, any state authority or any government body appears in our materials. Where we map against a published framework, such as the Australian Framework for Generative AI in Schools, we say "mapped against", and the mapping is available for your own review.
Our published accuracy evidence comes from established public writing-assessment benchmarks. It evidences how closely the engine agrees with expert human scores against holistic external rubrics; it is not a year-level calibration. No year level from 3 to 12 carries a year-specific accuracy claim. The product accepts Years 3 to 12; where year-level evidence does not exist, we say so rather than averaging it into a slogan.
Every published figure carries its sample, its confidence interval and its study limits. A single impressive number with the distribution hidden is the oldest trick in assessment marketing, and our own release checks block it.
No "most accurate", no best-in-category superlatives, no named-competitor superiority claims. There is no independent head-to-head benchmark in this category, so any such claim would be unverifiable. We publish our method and our numbers and invite you to ask every vendor for theirs.
Schools are being asked to trust generative artificial intelligence with something that matters: the fair assessment of a child's writing. That trust is earned by showing the working, publishing the limits, and building the product so that the teacher, not the tool, holds the pen on every official result.
These refusals are not modesty. They are what makes the claims we do make worth something: for every figure we publish, we can also tell you exactly what it cannot say, and we already have, one section above.