Skip to main content
FIG 1.0 — Rubric grading

Writing graded against the rubric, in seconds, not next week.

Students draft the profession-specific letter from case notes; the grading agent scores it across the official criteria and shows exactly where marks were won and lost — provisional until a teacher signs off.

FIG A — The agent

What the writing agent does

Scores the five criteria

Purpose, content, conciseness & clarity, genre & style, and organisation & layout — each scored with a short justification tied to your text.

Marks up your draft

Inline annotations flag irrelevant case-note detail, missed key information, register slips and structural issues, sentence by sentence.

Models a stronger version

The agent shows targeted rewrites of weak passages so the improvement is concrete, not abstract advice.

Tracks the pattern

Across attempts it surfaces the habits costing you marks — over-long sentences, copied case notes, weak openings — so teaching time goes where it counts.
FIG B — Format fidelity

How practice mirrors the sub-test

We match the structure, timing and on-screen behaviour of the official computer-based test so nothing on exam day is a surprise.

Task
One profession-specific task — typically a referral letter written from case notes.
Timing
45 minutes total — 5 minutes reading the case notes, 40 minutes writing.
Criteria
Scored across the official OET writing criteria on the 0–500 scale.
Surface
Typed in an exam-faithful editor; no copy/paste, authoritative state server-side.
FIG C — Feedback

What you get back

Every attempt produces structured, actionable feedback — provisional until a teacher signs it off.

  • A score per criterion with a plain-English justification.
  • Inline annotations on your own draft.
  • Suggested rewrites for the weakest passages.
  • A running view of recurring issues across attempts.

Honest note

Rubric scores are generated to mirror the official criteria for practice and are marked verified:false until a teacher signs them off. They are not official OET results and carry no certification.

FIG D — The full mock

Four sub-tests, one engine