Feedback given after a final grade is posted rarely changes a student's writing, because the assignment feels finished. Feedback on a draft, by contrast, can reshape an argument while there is still time to act on it. For essays on difficult works like Gorski vijenac, the gap between a first and a revised draft is often enormous. Building revision into the process is one of the highest-value changes an instructor can make.

The obstacle is time. Reading and commenting on a full set of drafts and then again on final versions doubles the grading burden, so many instructors skip the draft stage. AI-assisted feedback offers a way to provide timely comments on drafts without a matching increase in workload. The tool responds to the rubric you supply, so the guidance students receive matches the standards they will be graded against.

A practical model is a structured draft cycle. Students submit a complete draft by a set date, receive criterion-level comments within a day or two, and then revise before the final deadline. The instructor reviews the feedback samples to confirm quality and addresses specific cases in conferences. The student experiences a feedback loop that feels responsive and fair.

What good draft feedback includes

Effective draft feedback is specific, prioritized, and actionable. It points to the biggest issues first, such as a weak thesis or unexplained evidence, and leaves smaller matters of style for later. It also suggests a concrete next step that the student can carry out in one sitting. Comments that list every flaw overwhelm writers and lead to cosmetic changes instead of real revision.

• Identifies the one or two highest-impact problems first
• References specific sentences or paragraphs in the draft
• Explains why the issue weakens the argument
• Suggests a concrete step the student can take next
• Points out what is working so it is not lost in revision

Keeping the instructor in charge

AI feedback should supplement, not replace, the instructor's voice. Review a sample of comments each cycle, correct any that miss the mark, and adjust the rubric language if the tool misreads a criterion. Be transparent with students about how feedback is produced and encourage them to bring questions to office hours. Their trust in the process depends on knowing a human is accountable for the outcome.

Set boundaries so that the tool supports learning instead of doing the writing. Feedback should point to problems and prompt thinking, not rewrite paragraphs for the student. Explain that revisions must be the student's own work and that you will compare drafts to finals. These expectations preserve the integrity of the assignment.

Measuring the effect on learning

Track whether draft feedback improves outcomes by comparing the scores of first drafts and final versions. Look at which criteria improve most, since this reveals what students can fix with guidance and what needs more teaching. If argument scores stay flat, for example, you may need a mini-lesson on thesis development. The data helps you adapt instruction in real time.

Ask students for their view as well. Short surveys about which comments were most useful can show you where the feedback is clear and where it confuses. Over a few cycles, you can refine both your rubric and your process. The aim is a revision culture in which students expect to improve their work and know how to do it.