AI Platform

Meet the engines that do the grading.

GradeLoop isn't one model bolted onto a form; it's four specialized AI systems working as one pipeline, each obsessed with a single job.

Engine 01

Integrity Engine

Every submission, provenance-checked.

Before a single point is awarded, the Integrity Engine screens code for plagiarism and AI-generation signals, so the grade reflects the student's own thinking.

  • Similarity detection across Type 1–4

    Exact matches, renamed identifiers, restructured logic and semantic equivalents, caught even across sections and semesters.

  • Similarity clustering & side-by-side diffs

    Submissions are clustered by similarity type, so you can review every suspicious pair in a split-pane diff, with no manual cross-checking.

  • AI likelihood & plagiarism scoring

    A calibrated score estimates how much of a submission was machine-generated or copied.

integrity_report.jsonVERIFIED
Exact clone (Type 1)98%
Renamed clone (Type 2)91%
Restructured clone (Type 3)84%
Semantic equivalent (Type 4)76%

cluster · 3 submissions grouped as Type 2 · open side-by-side diff

provenance hash 0x8f3a…c21d logged

Engine 02

Grading & Tutoring Engine

Grades like an expert. Teaches like a mentor.

A rubric-aware LLM autograder evaluates logic, style and correctness, then a Socratic tutor turns every mistake into a guided discovery instead of a red cross.

  • Semantic autograder

    Understands intent, not just output, and awards partial credit the way a human grader would.

  • Socratic-style chatbot

    Whenever a student gets stuck mid-assignment, it responds with guiding questions that lead them to the answer, never the answer itself.

  • Instant grades & automated feedback

    On submission, work is evaluated automatically and detailed feedback with a grade lands in seconds.

Socratic Tutor · live sessionONLINE

Stuck on something?

Pick a question and watch the tutor guide, never tell.

Ask a follow-up…
Engine 03

Learning Analytics Engine

Authenticity you can audit.

Grades are only as trustworthy as the process behind them. GradeLoop links every submission to a verified human session with granular behavior analytics.

  • Isolated cloud coding platform

    Students work in sandboxed browser IDEs: every keystroke observable, nothing to install.

  • Biometric authentication & liveness

    Face verification confirms the right student is at the keyboard throughout the session.

  • Full session playback

    Replay any coding session keystroke-by-keystroke, a complete provenance trail from first line to final submit.

  • Struggle detection

    Failed runs, long stalls and hint requests are surfaced, showing exactly where and how long each student struggled.

session_timeline · student_2041MONITORING
Session started00:00

Face ID verified · liveness confirmed

Keystroke cadence04:12

Consistent typing rhythm, human-like

Struggle signal detected11:47

3 failed runs on edge cases, hint requested

Provenance check18:03

Code lineage traced to in-session edits

Session playback ready24:31

Full keystroke replay available for review

98%
Integrity
Low
Struggle
1
Flags
Engine 04

Oral Assessment Engine

Viva voce, without the bottleneck.

Oral exams prove understanding, but they don't scale. GradeLoop generates competency-based viva sessions from each student's own submission and evaluates responses automatically, for entire cohorts at once.

  • Automated assessment generation

    Questions adapt to each student's actual submission and target the competency it demonstrates, so no two vivas are alike.

  • Intelligent response evaluation

    Speech understanding scores depth of reasoning, not keyword matching.

  • Scalable verification

    Verify conceptual understanding across hundreds of students, consistently and fairly.

viva_session · q3 of 7LISTENING
AI Examiner

You chose a hash map over a sorted array. Walk me through the trade-off you made for memory usage.

evaluation · strong conceptual grasp · follow-up question generated on space complexity

response depth: highconfidence 92%
LTI 1.3 Native

Plugs straight into your LMS.

GradeLoop isn't a replacement for your learning management system; it extends it. Create once, link through LTI 1.3, and let the grades loop back on their own.

STEP 01

Create the assignment

Build your coding assignment in GradeLoop with rubric, starter code, test cases and constraints.

STEP 02

Link via LTI 1.3

Connect it to your existing LMS in one step. Rosters, sections and roles stay in sync automatically.

STEP 03

Students launch from your LMS

One click opens GradeLoop’s tracked browser IDE. No installs, no new logins, no friction.

STEP 04

Grades sync back

Scores and feedback flow straight into your gradebook, reviewable and overridable at any time.

Works withCanvasMoodleBlackboardBrightspaceAny LTI 1.3 platform
Under the Hood

Serious AI, serious outcomes.

Every piece of the stack earns its place: fairer grades, provable integrity, safe execution and a same-day LMS rollout.

Rubric-Aware AI Grading

reads logic & structure, not keywords

Cohort-Scale Plagiarism Detection

collusion caught across sections & semesters

Isolated Sandboxed Execution

student code never touches your infrastructure

Native LMS Integration

one-click rollout in Canvas & Moodle

See GradeLoop grade live.

Watch a real submission travel the full AI pipeline, from raw code to rubric score and Socratic feedback, in under two seconds.