Automate Peer Review Assignments with GPT-4.1-nano, Slack, and Email in n8n

Webhook receives student submissions → GPT-4.1-nano generates rubrics → Slack/email notifies reviewers → auto-scores and delivers reports

Who

Educators or course administrators managing collaborative student assessments

You get

Automatically assign peer reviewers, generate structured rubrics, aggregate scores, and distribute reports

You need

n8n · slack · email · postgres

Steps

1. Receive submission and assign reviewers

   Configure a Webhook node in n8n (path: peer-assessment, method: POST, response mode: responseNode) to accept submissions containing studentId, studentName, email, assignmentId, assignmentTitle, submissionUrl, and dueDate. Use a Set node to store the data, then a Code node to loop and assign 3 peer reviewers per assignment (circular modulo selection), outputting a reviewers array.

2. AI generates the review rubric

   Connect an AI Agent node (Generate Review Rubric) to an OpenAI Chat Model (gpt-4.1-nano) and a Structured Output Parser. Paste the source prompt into the Agent's text field and set the systemMessage from the source. Execution produces a structured rubric with 5 criteria (Technical Accuracy 25 pts, Problem-solving Approach 25 pts, Documentation Quality 20 pts, Code/Design Quality 20 pts, Innovation and Creativity 10 pts).

   Execution prompt · Source text。AI Agent node Generate Review Rubric text field (expression mode)

   Copy prompt

   =You are an expert engineering educator evaluating student assignments.
   
   Assignment Title: {{ $json.assignmentTitle }}
   Student Name: {{ $json.studentName }}
   Submission: {{ $json.submissionUrl }}
   
   Generate a comprehensive peer review rubric with the following criteria:
   1. Technical Accuracy (0-25 points)
   2. Problem-solving Approach (0-25 points)
   3. Documentation Quality (0-20 points)
   4. Code/Design Quality (0-20 points)
   5. Innovation and Creativity (0-10 points)
   
   For each criterion, provide:
   - Clear evaluation guidelines
   - Specific examples of excellent, good, and poor performance
   - Key questions reviewers should ask
   
   Total: 100 points
   
   Format the rubric in a clear, structured way that peer reviewers can easily follow.

   Execution prompt · Source text。AI Agent node Generate Review Rubric options.systemMessage field

   Copy prompt

   You are a helpful assistant that creates detailed, fair assessment rubrics for engineering students.

3. Notify reviewers

   Use a Slack node to post to a channel (e.g., #peer-reviews) with student name, assignment title, due date, reviewer list, and submission link. Simultaneously use an Email Send node to email reviewers with the assignment title as subject and the generated rubric embedded in the body (referenced via {{ $('Generate Review Rubric').item.json.output }}). Finally, use a Set node to prepare response data and a Respond to Webhook node to return a JSON confirmation.

   Execution prompt · Source text。Slack node Notify on Slack text field (expression mode)

   Copy prompt

   =📚 *New Peer Review Assignment*
   
   *Student:* {{ $json.studentName }}
   *Assignment:* {{ $json.assignmentTitle }}
   *Due Date:* {{ $json.dueDate }}
   
   *Assigned Reviewers:*
   {{$json.reviewers.map(r => `• ${r.reviewerName} (${r.reviewerEmail})`).join('\n')}}
   
   *Submission:* {{ $json.submissionUrl }}
   
   📋 Review rubric has been generated and sent via email.

4. Collect scores and calculate grades

   The Code node (Calculate Peer Score) currently uses Math.random() to simulate 5-dimension scoring and computes a total score and letter grade (A–F); replace with real peer-review data ingestion for production. Then store results in a peer_reviews table via a Postgres node. Use another Code node (Check Completion Status) to group by assignmentId and check if at least 3 reviews are collected; if complete, compute the average score and final grade.

5. Generate the final assessment report

   Connect an AI Agent node (Generate Final Report) to a second OpenAI Chat Model (gpt-4.1-nano) and Structured Output Parser 2. Paste the source prompt into the Agent's text field and set the systemMessage from the source. Inputs must include assignmentId, studentId, completedReviews, averageScore, and finalGrade. Output is a professional academic report with executive summary, strengths, improvement areas, actionable recommendations, comparison to class average (assumed 75/100), and next steps.

   Execution prompt · Source text。AI Agent node Generate Final Report text field (expression mode)

   Copy prompt

   =Generate a comprehensive final assessment report for the following peer review results:
   
   Assignment ID: {{ $json.assignmentId }}
   Student ID: {{ $json.studentId }}
   Completed Reviews: {{ $json.completedReviews }}
   Average Score: {{ $json.averageScore }}/100
   Final Grade: {{ $json.finalGrade }}
   
   Provide:
   1. Executive Summary (2-3 sentences)
   2. Strengths identified across reviews
   3. Areas for improvement
   4. Specific actionable recommendations
   5. Comparison to class average (assume 75/100)
   6. Next steps for the student
   
   Format as a professional academic report.

   Execution prompt · Source text。AI Agent node Generate Final Report options.systemMessage field

   Copy prompt

   You are an experienced engineering educator creating fair, constructive assessment reports.

6. Send reports and update dashboard

   Use an Email Send node to deliver the final report to the student, including final grade, average score, and report summary (referenced from Generate Final Report output). Use an HTTP Request node to POST grade data to an external dashboard endpoint (assignmentId, averageScore, finalGrade, completedAt). Then use a Code node (Analytics Report) to compute class average and grade distribution, and post the analytics to Slack with a final Slack node.

   Execution prompt · Source text。Slack node Post Analytics to Slack text field (expression mode)

   Copy prompt

   =📊 *Peer Review Analytics Report*
   
   *Total Assignments Completed:* {{ $json.totalAssignments }}
   *Class Average Score:* {{ $json.classAverage }}/100
   
   *Grade Distribution:*
   {{Object.entries($json.gradeDistribution).map(([grade, count]) => `• ${grade}: ${count} students`).join('\n')}}
   
   *Generated:* {{ $json.generatedAt }}

You are setting up an n8n peer-review automation workflow. Proceed in this order:
1. Import template workflow ID 10310 in n8n or recreate nodes from the provided workflow.json
2. Configure the Webhook node path as peer-assessment, method POST
3. Add an OpenAI API credential (model: gpt-4.1-nano)
4. Configure Gmail and Slack credentials
5. Replace the Math.random() logic in the Calculate Peer Score node with real peer-review data collection
6. Configure the Postgres database connection or substitute another storage node
7. Test a single assignment submission, confirming rubric generation, email delivery, and Webhook response
8. Before production deployment, verify all credential permissions and database table schemas