FAQ.

Edexia is an AI essay grading and feedback tool built for VCE English. It is trained by VCAA assessors, backed by Y Combinator, and used by schools across Victoria to provide detailed, criterion-level feedback on student essays.

Teachers upload or scan student essays. Edexia grades each essay against VCAA rubrics and generates paragraph-level feedback. Teachers then review, edit, and approve everything before students see it. Nothing reaches students without teacher approval.

In a trial across 579 VCE English essays at St Bernard’s College, teachers accepted Edexia’s grade 81.2% of the time and were within one grade band 98.3% of the time. This is consistent with examiner-level inter-rater reliability.

Edexia offers flexible pricing for schools. Contact us for details on plans and pricing for your department.

Edexia is used by VCE English departments at schools across Victoria, including St Bernard’s College. We work directly with Heads of English and VCE coordinators.

Yes. Edexia was incubated at Y Combinator, the accelerator behind Airbnb, DoorDash, Reddit, and OpenAI. The company is also backed by the founders of Clever, ClassDojo, and Coursedog.

Yes, when the system is trained specifically for the curriculum. Edexia achieves 81.2% exact agreement with teacher grades across 579 VCE English essays. Human examiners typically agree 60-80% of the time on subjective essays, placing Edexia at the upper end of human reliability.

Yes. Edexia supports every text on the VCE English study list. For each text, a knowledge base of themes, authorial intent, and key passages is maintained and validated by VCAA assessors.

Yes. Every criterion and grade descriptor from the VCAA English rubric is built into the system. Edexia evaluates essays against the same standards that VCAA assessors use during examinations.

Edexia is calibrated to VCAA rubrics for VCE English. Custom rubric support is on the roadmap, but the current focus is on delivering the highest accuracy within the VCE framework.

Generic AI tools lack understanding of VCE rubrics, text-specific knowledge, and assessment standards. Research from ETS found that GPT-4o scored approximately 0.9 points lower than human raters. Edexia closes this gap through VCAA assessor training, rubric calibration, and text-specific knowledge bases.

Yes. Edexia generates a draft grade and feedback that teachers review before release. Teachers can change any grade, rewrite any comment, or add their own feedback. Nothing reaches students until the teacher approves it.

Yes. Every comment Edexia generates is fully editable. Teachers can rewrite, delete, or expand on any piece of feedback. You can also add personal voice notes to any comment.

After Edexia grades a batch of essays, the teacher reviews each submission in the platform. Grades and comments are displayed side by side with the student’s text. You make any changes you want, then release feedback to students.

Yes. VCE English departments use Edexia to run moderation sessions. Teachers can review AI grades alongside their own, compare across the cohort, and calibrate as a group.

Edexia improves through ongoing moderation. As teachers review and adjust AI grades, the system learns school-level patterns and becomes more aligned with your department’s standards over time.

All data is siloed to individual school accounts, stored on Australian servers, encrypted, and de-identified before AI processing. Edexia never uses student data to train models for other schools. The platform holds SOC 2 Type II certification, ISO 27001, and ST4S accreditation.

Yes. Edexia is designed for use in Australian schools and complies with the Australian Privacy Act. Student data is stored locally, de-identified, and never shared or sold. Schools retain full ownership of all uploaded content.

No. Student essays are never used to train Edexia’s general models. All training data is siloed to the school’s account and is the intellectual property of the school.

Edexia holds SOC 2 Type II certification, ISO 27001, and ST4S (Safe Technology for Schools) accreditation.

Yes. Edexia includes AI detection tuned specifically for student writing. It uses writing replay (a keystroke-by-keystroke record of how the essay was written), paste detection, tab tracking, and behavioural analysis to identify AI-generated content with fewer false positives than output-scanning tools.

Most AI detectors analyze the finished text for statistical patterns. This approach produces significant false positives, especially for ESL students and formal writing styles. Edexia takes a process-based approach: it looks at how the essay was written, not just what the final text looks like.

Process-based detection is harder to circumvent than output scanning. Writing replay records the full composition process, so even if a student rewrites AI-generated content, the behavioural patterns differ from original writing. No detection system is perfect, but process data gives teachers substantially more information to work with.

Using AI as a first-pass feedback tool, where teachers review, edit, and approve all grades before release, is consistent with how schools already use technology in assessment. Edexia is designed so teachers always have the final say. The AI drafts; the teacher decides.

Edexia saves time on the mechanical parts of grading (identifying criterion-level strengths and weaknesses, drafting comments) so teachers can spend more time on the human parts (mentoring, conferencing, adapting instruction). Teachers who use Edexia report that they engage more deeply with student work, not less, because the time pressure is reduced.

Edexia’s assessor-trained models handle creative writing across the Creating Text framework. The system evaluates voice, narrative technique, and stylistic choices alongside structural elements. For edge cases or highly experimental work, teachers can override any grade or comment.

Edexia is trained by VCAA assessors who grade ESL student work as part of their regular practice. The system is calibrated to assess writing quality against VCE criteria, not against native-speaker norms. Schools should still review AI grades for any student cohort, which is why teacher review is built into every workflow.

That is a decision for the school and the teacher. Many schools using Edexia are transparent with students about the tool. Because teachers review and approve all feedback before release, the output students receive has been validated by their teacher regardless of how the initial draft was generated.

Edexia typically generates grades and feedback within 30 to 90 seconds per essay, depending on length. A full class set of 25 essays can be processed in under 10 minutes. Teacher review time varies, but most teachers report that reviewing and editing AI feedback takes a fraction of the time that grading from scratch would.

Yes. Edexia includes handwriting transcription that converts scanned handwritten work into digital text. Teachers can scan essays using the Edexia mobile app or upload photos. The transcription is reviewed alongside the original scan so teachers can verify accuracy.

Integration with learning management systems is on the roadmap. Currently, teachers upload essays directly through the Edexia platform.

Yes. Edexia is used for practice SACs, trial exams, and formative assessment throughout the VCE year. It gives every student detailed criterion-level feedback that would be difficult to provide by hand at scale.

Edexia starts with a department meeting to demonstrate the platform and answer questions. Individual check-ins with each teacher follow, then fortnightly touchpoints throughout the term. No formal training is required. The platform is designed to feel like moderating a colleague’s work.