Essay & short-answer grading
Grade short-answer and essay assignments at scale, with per-criterion feedback every student gets back. Especially valuable for high-enrollment intro courses.
Coming soon
Build open-ended response activities with Claude or ChatGPT, scored against your rubric.
Build open-response writing assignments in plain language. Learners answer in their own words, AI grades against the rubric you define, with per-criterion feedback and evidence pulled from their writing. SCORM-ready for any LMS.
What is it?
What are AI-graded open-ended responses?
An assessment format where the learner answers in their own words, an essay, a short answer, a case-analysis paragraph, instead of picking from a list. The AI grades each response against an instructor-defined rubric, with per-criterion scoring, evidence quoted from the response, and a specific next-step recommendation.
-
Free-text response interface, short answer, essay, structured-paragraph, code, or markdown -
AI grading against your rubric with per-criterion scores, not a single opaque grade -
Evidence pulled from the response so every score points to specific text -
Calibration mode lets you grade a few responses yourself, the AI matches your standards -
Exports as a SCORM 1.2 package for Cornerstone, Moodle, Canvas, TalentLMS, and every LMS -
Same bundle runs standalone in any browser for self-study without an LMS
How it works
How to build an AI-graded writing assignment
in four steps.
From a prompt and a rubric to a SCORM-ready open-response activity. No essay-grading service, no rubric XML.
01
Write the prompt
Describe what you want the learner to write. The prompt can include source material the learner should reference, or constraints on the response.
02
Define the rubric
Name the criteria you'll grade against, claim quality, use of evidence, structure, mechanics, with weights. The AI grades every response against the same rubric you'd hand a TA.
03
Calibrate (optional but recommended)
Grade 3–5 sample responses yourself. The AI matches your standards instead of inventing its own. This dramatically improves agreement between AI and human graders.
04
Export as a SCORM 1.2 package
One command produces a SCORM 1.2 zip. Per-criterion scores post back to the gradebook in any SCORM-compliant LMS.
Use cases by role
AI-graded responses
for every writing-heavy curriculum.
The same toolchain powers essay grading, short-answer assessment, code review, case-analysis grading, and reflective-writing assessment.
Constructed-response practice
Practice short-answer questions on demand. Students get rubric-aligned feedback immediately, so practice actually improves performance on the real exam.
Case-analysis assessment
Replace multiple-choice compliance quizzes with short-form scenario analyses. AI grades against the rubric, instructors review the borderline cases.
Code-explanation grading
Ask learners to explain code, debug an output, or write a short design-doc paragraph. Grade against rubric criteria, code clarity, root-cause accuracy, tradeoff articulation.
Reflection-journal grading
For professional development or clinical-practice training, grade reflections against rubrics like depth-of-analysis and evidence-of-application, not surface presence.
Performance-task scoring
For certifications that require constructed-response performance tasks, AI grading provides consistent, defensible rubric application across all candidates.
Rubric-graded responses
Every response graded against your rubric, criterion by criterion.
When the response comes in, the AI scores each rubric criterion separately. Each score points to specific text in the response, so both learners and instructors can verify exactly why a response got the score it did.
Update the rubric or recalibrate any time. Re-export the SCORM package and the next submission grades against the new rules.
Per-criterion scoring
Each rubric criterion gets its own score and feedback, not a single opaque grade. Learners see exactly where they need to improve.
Evidence-quoted feedback
Every score points to specific text from the response. Both learners and instructors can verify that the score reflects what was actually written.
Calibration mode
Grade a few responses yourself and the AI matches your standards. Dramatically improves agreement between AI grading and human grading.
Session complete
Your Debrief
9/10
★★★★★★★★★☆
Total 9 / 10
Claim quality: stated a clear, falsifiable position in the opening paragraph.
Use of evidence: cited at least two pieces of supplied source material and explained their relevance.
Counter-argument: acknowledged and addressed at least one defensible opposing view.
Plain-English authoring
Author AI-graded assignments in plain English. No rubric XML required.
Add a rubric criterion, change weights, raise the calibration sample size, or rewrite the prompt, just describe the change in your own words. Your coding agent rebuilds the grader and re-exports the SCORM package without you ever editing a rubric file.
When you're happy, export as a SCORM 1.2 package for your LMS, or use the same HTML bundle standalone.
Add a 'counter-argument' criterion worth 25% of the total.
Added criterion
counter-argument at weight 0.25. Other criteria re-balanced to sum to 1.Use these 4 sample responses to calibrate the grader.
Calibrated against 4 samples. Grader's agreement with your scores is now 92% on the holdout.
Export as SCORM.
✓ ai-graded-responses.zip ready (SCORM 1.2, 14 KB)
Assignment Library
Start from real open-response
assignments.
Paste any prompt into your coding agent to get a complete AI-graded assignment as a single self-contained HTML file. Adjust prompt, rubric, or calibration in plain language.
Higher ed
Intro-economics short essay
A 250-word short-essay prompt asking students to apply marginal analysis to a real-world example, with a 4-criterion rubric and 3 calibrated samples.
Prompt
Using /ai-graded-responses, build a 250-word marginal-analysis essay with a 4-criterion rubric. Calibrate against 3 sample responses.
K-12
Document-based question
A grade-10 history DBQ prompt referencing 3 supplied documents, graded on claim quality, document use, and counter-argument across a 6-criterion rubric.
Prompt
Using /ai-graded-responses, build a grade-10 DBQ on the New Deal with 3 supplied documents and a 6-criterion rubric.
Compliance
Scenario-based ethics response
A short-form ethics scenario asking the learner to defend a chosen course of action, graded on policy alignment, stakeholder consideration, and reasoning clarity.
Prompt
Using /ai-graded-responses, build a scenario-based ethics assignment with a 4-criterion rubric. Reward defensible disagreement with policy.
Engineering
Bug-write-up assignment
Engineers explain a debugging walk-through for a non-trivial bug, graded on root-cause clarity, evidence use, and prevention recommendation.
Prompt
Using /ai-graded-responses, build an engineering bug-write-up assignment with a 5-criterion rubric covering root cause, evidence, and prevention.
Clinical
SOAP-note grading
Trainee writes a SOAP note for a supplied case. Graded on subjective completeness, objective accuracy, assessment quality, and plan specificity.
Prompt
Using /ai-graded-responses, build a SOAP-note assignment from a supplied case. Rubric covers S, O, A, and P with weighted criteria.
Reflection
Mentee reflection journal
An open reflection prompt for a mentorship program, graded on depth-of-analysis, evidence-of-application, and forward-looking specificity, not surface presence.
Prompt
Using /ai-graded-responses, build a mentee reflection-journal assignment with a 3-criterion rubric rewarding depth, application, and specificity.
SCORM & LMS
SCORM 1.2 AI-graded assignments for any LMS.
Every AI-graded assignment exports as a standards-compliant SCORM 1.2 package. Upload the zip, assign it like any other course activity, and per-criterion scores flow back to the gradebook automatically.
-
Single SCORM 1.2 zip, upload to your LMS, no integration work, API keys, or custom JavaScript -
Completion, score, and time-on-task post back via cmi.core.score.rawandcmi.core.lesson_status -
Tested with Cornerstone OnDemand, Moodle, Canvas LMS, TalentLMS, Docebo, Brightspace, Absorb, 360Learning, SuccessFactors, and Workday Learning -
Same self-contained HTML bundle runs standalone on a public page, an intranet, or as an embedded iframe -
Optional integrity heuristics flag responses with AI-generated patterns and copy-pasted content -
Grading runs inside the bundle, no middleware, no data warehouse, no analytics SDK required
FAQ
Frequently asked questions about
AI-graded responses.
Rubrics, calibration, LMS compatibility, integrity, model choice, answered directly.
An assessment format where the learner answers in their own words, a short answer, an essay, a structured paragraph, and AI grades against an instructor-defined rubric. Each response gets per-criterion scores with evidence quoted from the response and a specific next-step recommendation.
Agreement rates depend heavily on rubric clarity and calibration. Without calibration, AI grading typically agrees with human grading at roughly the same rate as two human graders agree with each other. With calibration (you grade a few responses yourself), agreement goes up substantially.
Yes. Every per-criterion score is accompanied by evidence quoted directly from the response. Both learners and instructors can verify that the score reflects the actual writing.
The system flags borderline scores and responses with unusual patterns (very short, very long, off-topic) for instructor review. AI handles the long tail, instructors handle the judgment calls.
Yes. Every activity exports as a SCORM 1.2 package, which works in Cornerstone, Moodle, Canvas, TalentLMS, Docebo, Brightspace, and every other SCORM-compliant LMS. Per-criterion scores post back to the gradebook automatically.
Yes. Each assignment has its own rubric. You can reuse rubrics across assignments or write a new one each time, whatever fits your assessment design.
Optional integrity heuristics flag responses with AI-generated patterns and copy-pasted content. These are signals, not verdicts, the instructor still makes the academic-integrity call.
AI-Graded Responses is in active development. It will install via your coding agent with a single command, the same way the existing role-play skill works today. Add your email below and we'll let you know the day it ships.