§ Suite · ServicesFive tools · one workspace
What Illuma actually does.
Each service is one instrument. Pick the tab to see the tool, how it works, and what makes it different.
§ Under the hoodWhat the model actually measures
Every sentence gets analyzed against nine rubric dimensions. Sentences with specificity signals get one kind of mark; sentences where the essay is telling instead of showing get another. The margin comments show per-dimension scores (0-4) with the reasoning.
Output includes
- Per-dimension scores 0-4 with rationale
- Subject-check (intent vs actual)
- Paragraph-by-paragraph cohesion trace
- Craft moves checklist
Under the hood
ML models + calibration layers, not a ChatGPT wrapper.
1
The final score is computed, not guessed
Once the 9 dimensions are scored, the overall label comes from a deterministic formula, so the same essay always gets the same grade.
2
AI-detection flag
High AI-content signals get flagged so the evaluator can't be fooled by AI-assisted prose.
3
Prompt-aware grading
Different essay types are graded against different rubrics so a why-school essay isn't judged like an identity essay.
Ready to try it?
First eval free. No credit card required to sign up.