Nonrium AI
Understand the thinking.Support the next sentence.
Useful educational AI needs more than generation. Nonrium is designing a learning flow that connects criteria, evidence, human review and revision.
Design principles and development approach
What needs to sit behind a line of feedback.
We separate the roles of each step, from input to revision, and preserve the points where human judgment matters.
From assessment criteria to a better draft
Research & development designUniversity prompts · source texts · assessment criteria
Response analysis
Connect claims, evidence and structure to the rubric
Where is the evidence?Evidence-linked feedback
Locate the relevant passage and explain the assessment
Why this assessment?Instructor review
Check interpretation and add teaching context
What needs guidance?Rewrite
Compare revisions and shape the next learning step
What changed?
The design connects AI output to inspectable evidence, human review and a learner’s next draft.
Before a score, a reason to revise.
This is an educational illustration of our feedback approach, not a live AI request or a performance result.
Criteria
What the essay needs to do
Reflect university-specific prompts and criteria for reading, reasoning, evidence and expression in the feedback structure.
- Input
- Prompt · passages · assessment criteria
- Learning output
- A shared perspective for review
WRIDITIllustrative learning flowThe WRIDIT learning experience
Is cost enough to judge a public facility?
Passage A emphasizes shared public benefits. Passage B emphasizes efficient resource allocation.
Therefore, a costly public facility should not be built.
Distinguish the public-benefit criterion in A from the cost criterion in B.
Feedback that carries into the next essay.
Technology earns its value through learning.
We define the evaluation criteria first. Results will be shared once their scope and methodology have been verified.
Criteria → Observe → Review → Improve- Assessment consistency
- Compare educator assessment and AI feedback under the same criteria.
- Evidence relevance
- Review whether each reason to revise connects to the essay and the prompt.
- Learning behavior
- Measure feedback engagement, completed revisions and participation separately.
- Delivery economics
- Evaluate processing time, model cost and educator review time together.
Founder product experience
Experience turning AI into a product workflow.
Cofounder Kyoungsup Choi has planned ERP, CRM, proptech and legaltech products at Leysys. Experience translating complex rules into workflows, permissions and records informs Nonrium’s education product design.
Investment & partnerships
Help write the next chapter of essay education.
We welcome conversations with investors and partners interested in WRIDIT and online education.