Executive summary
An AP Literature and AP Language teacher grades each class set by hand, then runs the same essays through FRQuick as a second reader. When the comparison shows they applied the rubric too harshly, they sometimes raise a grade, and the final call stays theirs. Consistency is what earned their trust. General-purpose chatbots gave one essay different scores on different runs and gave essays of similar strength scores far apart, which makes calibration impossible. On the AP Lit and Lang rubrics, Row C rewards sophistication across the whole essay, so a closing line about human nature added after the analysis does not earn the point. Their two requests were feedback that ties each highlighted sentence to the rubric level it earned, which the results page now shows row by row, and a saved template for grading a whole class set.
In September we sent a short survey to people who had just signed up and graded a few essays. It asked three things: what's working, what's missing, and how they'd feel about a paid tier. The one reply we got came from a teacher who teaches both AP Literature and AP Language, and it was detailed enough that we asked to share it here. They agreed, as long as we left their name out. Their reason is at the bottom of this post.
Everything below comes from that one email, and we quote the teacher directly wherever the words are theirs.
How does a teacher use an AI essay grader?
This teacher grades a class set by hand first. Only after that do they paste the same essays into FRQuick, because it "will occasionally highlight things I myself overlooked."
Sometimes the second read changes a grade. If FRQuick convinces them they were too harsh in applying the rubric, they raise the score, and the grade that goes in the gradebook is still one they decided on. What they get from the tool is a second opinion that shows its work, which is how we'd want any teacher to use it.
That setup is close to how AP scoring works each June. Readers train on one rubric, score thousands of essays, and get checked against each other, since two careful people can read the same paragraph and land a point apart. If you've never seen that process, how AP readers score essays walks through it.
Why did consistency matter so much?
They had already tried general-purpose AI chatbots for the same job. Two problems kept showing up. The same essay would come back with different scores on different runs, and two essays of about the same strength would get scores far apart. A teacher can't calibrate against that kind of drift.
Their verdict on FRQuick was that it "provides the best balance a computer could in not being overly lenient with the rubric yet still rewarding the student for what they've done well."
Most of our calibration work went into that balance, because a grader that hands out 5s feels encouraging for about a week, right up until a practice exam comes back scored by a person. Our benchmark results show how our scores line up with human AP readers. For what those numbers can and can't tell you, read is AI accurate at grading essays?
What does it take to earn the sophistication point?
Then came Row C. The teacher liked that FRQuick "doesn't flippantly award the sophistication point for a sentence or two that's added as an afterthought and not integrated into the student's argument."
Students lose this point constantly. Plenty of essays close with a big line about human nature or society, and the writer assumes that line counts as complexity. It doesn't. The College Board's scoring guidelines say the point has to come from the essay as a whole, and a single phrase or reference isn't enough to earn it.
Essays that earn it usually carry a single idea through the whole argument. Picture a tension in the passage, say a narrator who sounds calm while describing something frightening, that each body paragraph complicates a little further. A closing sentence about the human condition, added after the analysis is already over, can't do that. The AP essay rubric, explained breaks down all three rows, and our guides to the prose analysis essay and the rhetorical analysis essay show what Row C looks like on those specific tasks.
What did they ask us to change?
Their main request was for students. FRQuick already marks strong and weak sentences and labels each comment with the rubric row it belongs to. They wanted each comment tied back to the score in the rubric's own language, closer to "this sentence in paragraph 3 is why you got 3 out of 4 on Evidence and Commentary." AP rubrics are formulaic, they pointed out, and a breakdown of where a student's writing sits inside that formula "would help immensely."
We agreed. It was the first change we made after reading the email. On your results page you can now click any row of your score. It opens a short explanation of what that score level means on the rubric, what the next level up asks for, and every comment tied to that row, numbered in the order they appear. Click one and the essay scrolls straight to the sentence.
The second request was about their own workflow. Grading a class set means choosing the essay type, pasting the prompt, pasting the passage, and then pasting each essay, over and over for every student. They asked for a saved template so they'd only have to paste the new essays, and we're now building a batch mode for teachers around exactly that idea. For how teachers use FRQuick right now, and what it deliberately doesn't do, see the teachers page.
Why they asked us to leave their name off
Teachers spend a huge amount of energy keeping students from using AI on their assignments. If a student found this teacher's name on an AI company's website, they expect "an instant barrage of whining and complaints about how hypocritical it is." They see a real difference between a teacher using AI to double-check their own grading and a student using it to skip the work. They also don't expect a room of sixteen-year-olds to give that difference much credit.
We think they're right, and students should draw the same line with FRQuick: use the feedback to see why a paragraph scored the way it did, then go rewrite the paragraph yourself.
If you want to try that on your own writing, grade a practice essay with the AP Lit grader or the AP Lang grader, then click the Evidence and Commentary row to see which of your sentences it points to. For comparison, the student essay library has scored essays from other students, each with notes on what earned its points.
FRQuick is not affiliated with the College Board or Advanced Placement. AP is a registered trademark of the College Board.
Frequently asked questions
How can an AP teacher use an AI essay grader?
As a second reader after grading by hand. One AP Lit and Lang teacher scores a class set first, then runs the same essays through FRQuick to catch details they missed, and occasionally raises a grade when the comparison shows they applied the rubric too harshly. The teacher still decides every final grade.
Why does scoring consistency matter in an AI essay grader?
A teacher can only calibrate against a score that holds still. General-purpose chatbots can return different scores for the same essay on different runs, and very different scores for essays of similar strength. A grader tied to one fixed rubric should give the same essay the same score every time it is graded.
What earns the sophistication point on the AP Lit and Lang rubrics?
Sophistication shown across the essay as a whole. The College Board's scoring guidelines say a single phrase or reference does not earn Row C, so one closing sentence about human nature or society added after the analysis rarely counts. Essays that earn it usually develop one tension or complexity through every body paragraph.
Can AI feedback show why an essay got its rubric score?
Yes, on FRQuick's results page. Clicking a row of the score, such as Evidence and Commentary, shows what that score level means on the rubric, what the next level up requires, and every comment tied to that row. Clicking a comment scrolls the essay to the sentence it refers to.
Is it hypocritical for a teacher to use AI while banning it for students?
The two uses differ. A teacher using AI to double-check their own grading keeps the judgment and does the work, while a student using AI to write an essay skips the practice the assignment exists for. A student who reads AI feedback, finds a weak paragraph and rewrites it themselves is using it the way the teacher does.
