Honest answers
No. Not our AI, not anyone’s. We build an AI admissions tool and this page is us telling you that plainly. Real decisions depend on the rest of that year’s pool and on needs inside the college that never show up in public data. What AI can honestly do is different, and more useful than a made-up percentage: it can show you how a committee would argue about your file.
01 · Why the answer is no
None of these are AI limitations that the next model release fixes. They are facts about how admissions decisions get made.
01
A committee is not grading your file against a fixed rubric. It is choosing a class from the applicants who showed up that year. Your file can be identical two years running and get two different answers, because the pool around it changed. No model sees next year’s pool. It does not exist yet.
02
Colleges manage yield, program capacity, budget, and enrollment targets that never appear in any public dataset. A school that quietly needs oboists, or more engineering deposits, or fewer deferrals from one region, will make decisions no outside tool can anticipate. This is not a data-quality problem that better AI fixes. The information is simply not published.
03
Recommendation letters, interview reports, and school context forms shape real decisions, and no consumer tool has them. Anything claiming to predict your outcome is reasoning from a partial copy of your application. That can still be useful. It cannot be complete.
04
When a school admits well under a tenth of applicants, most qualified files get denied. That is arithmetic, not pessimism. A tool that flatters you at a reach school is not being generous. It is being wrong in the direction that sells subscriptions.
02 · Two different machines
Chancing calculators are not useless. They are a different machine than the one that decides your application, and the gap between the two is where students get misled.
| A chancing calculator | A committee-style read | |
|---|---|---|
| GPA and test scores | Yes. This is most of what they run on. | Yes, anchored to what the school itself publishes about its enrolled class. |
| Course rigor in context | Sometimes, as a self-reported checkbox. | Read against what your school offers and what you chose. |
| Essays | No. A percentage cannot read. | A specialist reads them and files a report before the officers weigh in. |
| Whether the pieces agree with each other | No. Each input is a separate field. | Yes. An essay that contradicts the activity list gets called out in writing. |
| The disagreement | Hidden. You get one number. | Recorded. Officers disagree in writing and a chair resolves the split. |
| Your exact odds | Claimed, with false precision. | Not claimed. A simulation is a rehearsal, not a forecast. |
The right column describes Trajecta’s Committee Review. It is a simulation, and simulations have limits too. We list ours on the committee page.
03 · The honest alternative
You cannot know your odds. You can know what the conversation about your file would sound like, and that is the thing you can actually act on.
Trajecta’s Committee Review puts your entire application in front of a simulated admissions committee. A specialist reads your essays first and files a report. Then four officers read the whole file through different lenses, academics, narrative, impact, and context, and react to each other in writing. They disagree. The disagreement is recorded, and a chair resolves it into a decision with next steps.
The reads are anchored to the school’s own published numbers, the Common Data Set that colleges file about their enrolled class, instead of folklore about what a school wants. And the output talks the way a room talks. Not “predicted: reject” but “this room denies files like this most cycles, and here is what would change the conversation.”
We published a full run on a real application, end to end, four reads and a split vote and the decision, so you can judge the thing yourself before signing up for anything.
04 · Practical advice
05 · FAQ
No. Real decisions depend on that year’s applicant pool and on institutional needs like yield and program capacity that no outside tool can see. AI can estimate where your record sits against a school’s published numbers, and it can simulate how a committee might argue about your file. Neither of those is a prediction of your outcome, and any tool that promises one is overclaiming.
Accuracy for a single applicant is close to unknowable, because you only apply once and never see the counterfactual. Tools can be directionally right about bands, meaning they can usually tell a likely school from a reach. Treat any tool that advertises a precise personal percentage at a selective school with suspicion, whatever its accuracy claims.
A predictor outputs a number and hides the reasoning. A simulation stages the reasoning and skips the false number. Trajecta’s Committee Review has four AI officers read your entire application through different lenses, disagree in writing, and end in a chair’s decision with next steps, phrased the way a room talks about files like yours. You leave knowing what the argument about your application would be, which is something you can act on.
For sorting a long list into reach, target, and likely bands, yes, they are a reasonable starting point and some good ones are free. For deciding how your actual application reads, no. We wrote up the accuracy question separately, including what calculators are genuinely good at.
Two honest layers. The college list scores fit and admission probability from each school’s published data, as bands rather than promises. Committee Review then rehearses the finished application: an essay specialist files first, four officers read the whole file and react to each other, and a chair resolves the recorded split into a decision and next steps. Calibrated against real, publicly shared admissions outcomes. Never sold as a predictor.
Build your profile free, then watch a committee read a real application end to end before you decide anything.