AI-assisted assessment for teachers and auditors

The AI grades first. You decide.

Upload any kind of work — reports, code, scanned exams. EvalFlow marks it against your rubric, cites the slide it came from, and hands you a queue to review, override and publish. Every change is logged.

Used at 4 universities and 1 audit firm · 1.2M pages graded · reviewer agreement 0.91

✦ AI first pass · teacher decides
Q2 · Active scanning · AI 6 / 10 · conf 0.83Nmap flags are listed but the rationale for -sS vs -sT is missing. Cites: Slides W4 p.6–8.
045678override · reason logged
−58%grading time per paper
0.91AI ↔ reviewer agreement
96%papers with an AI first pass
7 daysappeal window with evidence
Trusted byMahidol UniversityKasetsart UniversityKMUTTA&C Audit PartnersForti5 Tech
One engine · three industries

The vocabulary follows your work.

The same pipeline — receive, AI first pass, human review, publish, appeal — but screens, policies and words change with your tenant's industry.

For teachers

Mark 200 papers in an afternoon — with a reason on every mark.

Reports, code, notebooks, scanned exams. The AI marks against your rubric and cites the slide; you review with the keyboard and publish with an appeal window.

  • Design assessments from your own slides
  • Grade bands with auto-tuning
  • Calibrate teaching assistants
  • Thai-language student portal
CourseAssessmentStudentGrade
CSS484 · Ethical Hacking2026-1 · 58 students
15to grade
0.91agreement
3appeals
B+median
Assignment 2 — Reconnaissance reportGrading
Lab 3 — Port scanner (code)AI pass done
Midterm — scannedPublished
Quiz 5 — onlineLive
How it works

Four steps from submission to published result.

  1. 01Receive

    Upload, scan, link a repo, or let participants submit. Text, code and answers are extracted automatically.

  2. 02AI first pass

    Marked against your rubric or criteria, with a reason, a confidence and a citation for every mark.

  3. 03Human review

    A keyboard-first queue: accept, override or comment. Every change is logged with a reason.

  4. 04Publish & appeal

    Results go out with an evidence-backed appeal window. People see the reasoning, not just a number.

Capabilities

Built for real assessment, not a chatbot.

Everything a teacher, auditor or interviewer needs in one working day.

01Every kind of work

Reports, code run in a sandbox, notebooks, scanned exams with MCQ, fill-in, numerical and written answers.

02Design assessments with AI

Grounded in your uploaded slides and handouts; every item cites its source page.

03Keyboard-first review

J/K to move, digits to mark, A to accept the AI. Three minutes a paper.

04Calibration

See which grader drifts on which question, and fix it before publishing.

05Workflows you can edit

Canvas or YAML, with human gates, retries and version history.

06Transparent to students

Per-question feedback with slide citations, and an appeal path that reaches a human.

Trust

AI proposes. People decide. Everything has evidence.

Every mark is explainable, appealable and auditable — by design, not by policy document.

No automatic decisions

The AI proposes a mark and a reason; a person always confirms. Low-confidence marks are routed to review.

Evidence at every step

Submissions, evidence and marks are hashed and timestamped. Published results can't change without a logged reason.

Data stays in Thailand

Hosted in-country, PDPA-ready: consent capture, subject access and erasure requests built in.

Fully auditable

Every action, role change and PII access is in the audit log — exportable for your own auditors.

“Two weeks became two afternoons — and every student got a reason, not just a number.”
Dr. Prasert W.Lecturer · Computer Engineering
“First engagement where auditees uploaded their own evidence and we signed off without chasing email threads.”
Arunee S.IT audit lead
“Panellists only see each other's scores after submitting. Debriefs are half as long and decisions are clearer.”
Thana R.Head of talent
Case study

Cut grading time 58% without cutting corners

A 74-student Ethical Hacking course ran a full term on EvalFlow: AI first pass on every paper, two TAs reviewing, one teacher publishing.

Read the case study
Pricing

Start free. Pay for the AI you use.

Unlimited seats · AI credits cover design, grading and extraction · Thai data residency on every plan.

StarterFree

1 course or engagement · 300 AI credits / month · unlimited users

Start free
Team฿4,900 /month

Unlimited courses · 5,000 credits · workflows · integrations · Google Classroom, Slack, Discord

Start 14-day trial
InstitutionLet's talk

Multiple faculties or units · SSO · retention policies · SLA · named account manager

Contact sales
FAQ

Questions we get asked.

No. It proposes a mark with a reason; a teacher or reviewer confirms, overrides or rejects. Nothing is published without a human decision.

Grade your first batch today.

Upload ten papers and get a first pass in five minutes. No card needed.

© 2026 EvalFlowProductPricingContactPrivacyTermsEnglish · switch to TH above