Most AI writes.
Illuma predicts.
An ML model that reads your whole application, ranks every candidate move by expected admit-rate delta per school, and shows what to do next.
Sign up and start your 14-day free trial. No card required.
The data behind every prediction.
Thousands of real, de-identified admissions cycles feed the model. Browse the corpus directly — every profile below is what the similarity engine sees when it ranks your peers.
Activities in the corpus
Real profiles, real decisions
Where to use Illuma
Essay evaluation
9-dimension scores with rationales. Catches the subject/story gap, cohesion holes, and missing craft moves before submission.
Admissions Compass
What to do next, ranked by impact. The Compass runs the same ML that scores your admit chances across candidate moves — raise SAT, add APs, strengthen ECs — and shows the delta in probability points per school.
Supplement brainstorm
Five candidate directions for a school's supplement, anchored to specific moments in your background.
Why not just Grammarly, ChatGPT, or an essay coach?
| What matters | Grammarly | ChatGPT | Essay coach | Illuma |
|---|---|---|---|---|
| Predicts admit probability per school from your profile | × | × | × | |
| Admissions Compass — ranks your next moves by admit-rate delta | × | × | ~ | |
| ML calibrated on real historical outcomes (~250 schools) | × | × | × | |
| Retrieves peer essays from similar profiles | × | × | ~ | |
| Personalized to your spike, major, activities, and stats | × | ~ | ~ | |
| Archetype-aware — knows how your pool is being read | × | × | ~ | |
| Grounded in per-school CDS data (~250 schools) | × | × | ~ | |
| Same methodology every time (reproducible) | × | ~ | ||
| Feedback in seconds, not days | × | |||
| Cost per revision | $$$ |
See the full write-up
Grammarly
Catches grammar, tone, clarity.
- ·Doesn't know what 'specificity' means for a college essay.
- ·Doesn't catch a duplicated paragraph that wastes 80 of 400 words.
- ·Doesn't tell you the prompt is asking for vulnerability and you wrote a credential list.
- ·Doesn't know what a Tier-1 anchor looks like at a top-30 school.
Per-dimension scoring on the same 9-axis rubric admissions readers use — plus the Admissions Compass runs an ML model that predicts your admit probability per school and ranks candidate moves (essay upgrades, test retakes, activity work) by expected admit-rate delta. Paragraph-level cohesion trace + subject-vs-story diagnosis on top.
ChatGPT
Generates fluent prose on demand.
- ·Different answer every session — no consistent rubric or memory of your profile.
- ·Confidently invents school facts (a professor's research, an admit rate, a program name).
- ·Doesn't know which essay shape an AO has read 5,000 of and which is rare.
- ·Will rewrite your essay for you — and Claude/GPT prose is exactly what AOs flag as AI-assisted.
Grounded in counselor methodology + a real ML model calibrated on ~250 schools' historical outcomes. Consistent every time, anchored to YOUR profile. The Admissions Compass shows which move (essay revision, higher SAT, exceptional-tier activity) would move your probability the most at each school on your list.
Essay / writing coaches
Personal feedback from someone who's read essays before.
- ·$150-$500/hr; full packages run $5K-$25K+. Most families can't afford this.
- ·Quality varies wildly — many coaches haven't actually read for a top admissions office.
- ·Slow — you wait days for feedback on each revision.
- ·Only works for the schools they happen to know well.
Counselor methodology + predictive ML model built in — every essay eval feeds back into the Admissions Compass, which predicts your admit probability per school and ranks candidate moves. Replies in seconds, costs cents per run, covers ~250 schools' CDS data uniformly, and works across every school on your list at once.