Help Center
AI Tools6 min read

Understanding AI feedback

What your essay grade, grammar check, and cliché scan actually measure — and how to act on them.

Updated July 2, 2026


Draftl gives you three kinds of AI feedback on any essay: a full grade, a grammar check, and a cliché scan. All three live in the tools panel on the right side of the essay editor, each in its own tab. This guide explains what each tool measures, how to read its results, and how to turn them into a stronger draft. If you're still setting up your first essay, start with writing your first essay.

The Essay Grader

Open the Essay Grader tab (the graduation cap icon) and click Grade My Essay. Draftl scores your draft across six dimensions, each from 0 to 100:

DimensionWhat it measures
AuthenticityWhether the essay sounds like a real person's genuine experience
ConcisenessHow efficiently you use your words — no padding or repetition
FlowHow smoothly ideas and paragraphs connect
HookHow well your opening grabs the reader
UniquenessWhether your topic and angle stand out from typical essays
VoiceHow distinctly your personality comes through

Your overall score is the average of the six, shown in a large circular meter, with each dimension in its own smaller gauge below. Alongside the scores you'll get 2–4 strengths and 2–4 improvements as short bullets, so you know both what's working and what to fix first.

Here's how to read the overall number:

ScoreMeaning
90–100Exceptional
80–89Strong
70–79Good
60–69Average
50–59Below Average
Below 50Poor

Tip: Most essays land between 60 and 80 — that's normal, not a red flag. A first draft in the 60s with clear improvement bullets is exactly where the process is supposed to start.

After you revise, click Grade Again to see how your changes moved the numbers. Focus on your two lowest dimensions rather than trying to raise everything at once.

Grammar Check

The Grammar Check tab (the spell check icon) handles the mechanical layer. Click Check Grammar and Draftl scans your draft for issues in six categories: grammar, spelling, punctuation, style, mechanics, and clarity.

Each issue appears as its own error card showing:

  • The exact sentence where the problem occurs
  • The rule being broken
  • A suggested correction
  • A plain-English explanation of why it matters

Click any card and Draftl highlights the sentence directly in your editor, so you never have to hunt for it. If you disagree with a suggestion — style calls are often judgment calls — you can dismiss the error, and restore it later if you change your mind. Run Check Again after edits to confirm you're clean.

Cliché Detector

Admissions readers see the same phrases thousands of times, and the Cliché Detector tab (the repeat icon) exists to catch them before they do. Click Find Clichés and Draftl flags:

  • Overused phrases like "changed my life" or "outside my comfort zone"
  • Tired tropes, such as the big-game win or the mission trip revelation
  • Generic filler words like "passionate" and "well-rounded"
  • Predictable openings, including famous quotes and rhetorical questions

Each finding gets a severity level — severe, moderate, or mild — with the most serious issues listed first and an overall severity gauge up top. Every finding explains why the phrase falls flat and gives a concrete suggestion for what to write instead. Like grammar errors, findings are clickable to highlight the phrase in your editor, and you can dismiss or restore them.

Tip: A "severe" cliché is usually a topic-level problem, not a wording problem. If your central story gets flagged, consider a brainstorming session with the Writing Agent to find a fresher angle.

Putting the three tools together

The tools work best in a deliberate order. Start with the Cliché Detector while your draft is still rough — it's cheapest to swap out a tired phrase or rethink a predictable opening before you've polished the surrounding paragraphs. Next, use the Essay Grader to judge the essay's substance: story, structure, and voice. Save the Grammar Check for last, once the content has settled, so you're not proofreading sentences you're about to rewrite anyway.

After each round of feedback, make your edits in the editor and re-run the relevant tool. Watching your Hook score climb or your severe clichés disappear is the clearest sign your revisions are landing.

Choosing an AI model

The model selector in the top right of the editor controls which AI model powers your feedback. Free plans can choose between Gemini 3.5 Flash, Claude Haiku 4.5, and GPT-5.4 mini. Pro unlocks the full lineup, adding Gemini 3.1 Pro, Claude Sonnet 4.6, Claude Opus 4.8, GPT-5.4, GPT-5.4 Pro, DeepSeek V4 Pro, Grok 4.3, and Kimi K2.6. If you're not sure, the default model is a solid choice for all three tools.

Free plan quotas

Feedback tools have monthly limits on the free plan: 1 essay grade, 5 grammar checks, and 5 cliché checks. Quotas reset at the start of each billing cycle, and you can track them on the Usage page. Since your free grade is precious, run it after a serious revision pass rather than on a rough first draft. Pro makes all three unlimited — see managing your subscription for details.