Most AI guesses.
Illuma looks it up.
Every prediction is grounded in a large corpus of real, de-identified admissions cycles. And the corpus is yours to search.
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. Peek at a slice below, or sign in and ask the chat: 'what % of sub-1500 SAT applicants got into Stanford?'
Real profiles, real decisions
How the database worksWhere to use Illuma
Applicant database
Search a large corpus of real, de-identified applicants. Stats, activities, every school decision. Ask the chat in plain English: 'what % of sub-1500 SAT applicants got into Stanford?' and get real aggregates.
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.
Essay evaluation
9-dimension scores with rationales. Catches the subject/story gap, cohesion holes, and missing craft moves before submission.
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.
We never make things up.
Students trust Illuma to guide real decisions about their time, their money, and their future. That trust is fragile. So we make a promise: we do not present information we cannot verify, we do not cite sources that do not exist, we tell you when we do not know, and when we get something wrong we fix it publicly. Accuracy is not a feature. It is the ground everything else stands on.