A third fewer mistakes in one term.
For schools and learning centers. Gaplight finds every student's gaps before the exam does and shows each teacher what to teach next. If mistakes don't fall by a third, the next term is on us.
Students already put in the hours. Gaplight sends those hours to the gaps that cost points. This is what that did for our first cohort.
- −50%
- mistakes for the median student, first attempt against last
- 71%
- of students who started below 75% readiness gained 15 points or more
- 51%
- reached 90% mastery of a topic within 11 days
Gaplight, first cohort: 45 students preparing for Ukraine's national exam (NMT), September 2026.
Introducing the Gaplight engine
Six steps under every screen. Every answer runs through all six.
Every skill the exam measures.
Weighted by how much of the section it carries. SAT is built from the College Board domain map.
The Gaplight guarantee
Or the next term is on us.
Your cohort gets at least a third fewer questions wrong by the end of a term, or the next term is on us.
Mistakes are counted on fresh attempts at the exam's own question bank, across the whole cohort, first attempt against last. It is the number we show you every week, so you can check the promise long before exam day.
Why a third: in our first cohort the median student made half as many mistakes by the end of a topic. We promise less than we measured.
We do not guarantee a number on the official exam. No one can, and the research on commercial test prep does not support the figures the category advertises. We guarantee the thing we measure — and we show you the measurement.
Every other tool answers to the student. We answer to you.
Teaches your student without you
Hands your teachers the next hour of work
Turns any file into flashcards
Ranks every gap by what it costs on the exam
Guarantees a number on exam day
Guarantees the number we measure for you weekly
Reports progress to the learner
Reports readiness to the school
students on the largest free prep platform never reach six hours of practice
points — what peer-reviewed research credits commercial coaching with on the SAT, against a standard deviation above 100
answers logged one by one on the exam the engine runs on today, every one traceable to a single student and a single skill
Gaplight, cohort data, September 2026
What your team sees on day one.
One model of each student sits under every screen, so the forecast, the plan and the report always agree.
Class forecast
1296
± 30 · +48 since Sep
On track for target
17 / 28
+4 this month
Skills below target
12 / 29
class median
Biggest gap
Nonlinear functions
17 students · +31 pts at stake
Skill map by student
level 1–7 · 8 SAT domainsNeeds a conversation
Off pace for their target by the test date.
- Owen P.1080 → 12009 wks behind
- Tom W.1110 → 12005 wks behind
- Daniel K.1210 → 12502 wks behind
Updated after every session
Every student on one map. The whole class across the eight SAT domains, the forecast for each student, and the gap most of them share.
See every screen →Names and numbers on these screens are illustrative.
Every classroom. Every center.
One engine under the people who prepare students for the exam.
Know where the whole cohort stands.
Diagnose a year group in one session. A plan per student, a view per class, a report for the board.
Gaplight for schools →Student data stays with the school.
Schools and centers hand us their students' work. These are the rules we hold ourselves to. Read the Privacy Policy
- Schools own itRecords belong to the school or center that enrolled the student.
- Never soldNo selling student data, no advertising built on it.
- Only what the plan needsAnswers and progress. No personal profiles.
- Deleted on requestA student, a class or a whole account, whenever you ask.
- Signed firstA data agreement with every school before the first login.
Founding pilots · 2026–27
Test it on your own students first.
We are new and choosing our first schools and centers. Ten weeks with one class, free, with the measures agreed before it starts.
- Week 0Student list, sign-in, data agreement
- Week 1Diagnostic and first forecasts
- Weeks 2–9Practice, gap ranking, session plans
- Week 10Full practice test and a review against the agreed measures
See Gaplight on your students' exam.
A walkthrough of the diagnostic, the plan and the teacher view, and a pilot plan sized to your school or center.
Questions schools ask first.
Which exams does Gaplight cover?
The SAT is fully built today. The engine runs on any exam with a clear skill map; tell us which exam you teach to and we will show you where it sits on our roadmap.
How is this different from a question bank?
A question bank gives students more questions. Gaplight decides which question comes next, based on which gaps cost the most points, and tells teachers what to teach.
Does it replace tutors or teachers?
No. It makes their time precise: who needs help, on what, and what to plan for the next session.
How long does it take to start?
About an hour of setup: a student list, a call and the data agreement. A class can take the diagnostic in its first session.
How is it priced?
Per active student per season, with plans for learning centers, schools and districts. Every plan starts with a free ten-week pilot. See the pricing page, and we size it with you on the demo.
Does it work with Claude or ChatGPT?
Yes. Gaplight has its own MCP server. Once the school turns it on, students and teachers connect it to Claude or ChatGPT with a personal token and work with their own data in plain language.
What happens to student data?
It belongs to the school or center, is never sold, and is deleted on request. Every school signs a data agreement before students log in.