For US public charter schools. Founding partners. Nothing measured yet.
Four students scored 62 on the same interim. Nothing in that report tells a teacher what to do on Tuesday.
PrepGraph shows a teacher, before first period, which students are stuck on which concept and which concept to reteach first. We have not measured a learning outcome and we will not claim one. See the gap. Name the students. Re-check the same list later — and publish what that does and does not show.
Constructed to show the shape of the output. Four students in one seventh-grade section score 62 on the same fall benchmark. PrepGraph never sees that test and never scores it — the 62 is yours. What PrepGraph has is each student's own work inside PrepGraph on the same concepts, recomputed overnight. All four are entered as approaching; when the intervention block is scheduled, all four are sent to the same group and handed the same worksheet.
In her PrepGraph practice she picks the same pre-tagged wrong answer on sign-distribution items: a minus sign in front of a parenthesis gets distributed to the first term and not to the rest. That comes from which specific distractor she chose in our items, not from any score.
What it would takeTen minutes on one habit, this week, before it becomes the method she trusts. She reads struggling on that one concept and understand on the rest, so she sits at the top of the reteach queue there and nowhere else.
In his PrepGraph work the arithmetic is clean and the setup is not: he reads understand on the computation concepts and struggling on translating a written situation into an equation.
What it would takeSomething other than more math practice. The reteach queue puts him in a different place from Student A, and the thing that stopped him will stop him again in science.
Every concept PrepGraph has evidence for reads understand. She read each question twice in a language she does not speak at home and answered the last two in fragments because the clock ran out. Reading under time is not something this software sees.
What it would takeNothing from us. The honest output is to say we have nothing to add rather than manufacture a gap. The grid leaves her out of the reteach group; whether that is worth anything to your school is your judgement, not a finding of ours.
He has not used PrepGraph. His row is blank — unmeasured, not zero, and never green by default.
What it would takeUsage, or nothing. If most of your roster looks like Student D, the grid is empty and there is nothing for us to report. That is the most common way a pilot like this dies, and it is the one thing we cannot fix for you from a distance.
One number, four students, four different actions — and two of the four are not us. Two of them do not need more math, and one of them we have nothing to say about at all. The intervention block hands all four the same worksheet because nothing in the report separates them, so nothing in the schedule does either.
Four people have to agree
What this is, from where you sit
The answer is different for each of you, and none of them is a growth number, because we do not have one.
Executive Director / CEO
We will not hand you our number. We will help you produce yours.
We have not measured a learning outcome — no study has begun, no result exists — and we have measured nothing about enrolment, retention or family satisfaction either. What we offer is an instrumented pilot on a baseline instrument you choose and we do not score, with a written commitment that whatever it shows gets published, including if it shows nothing. Two commercial facts before anything else, because they are the likeliest reason we cannot do business this year: there is no US school price in our billing system at all, and our code refuses school-level access grants as ineffective, so today the only mechanism that keeps a student's access on is a family buying a seat. We will not run that in a US public school — we will not use a roster you gave us to sell to your families. That means either we have built a school-paid seat and shown you the invoice line, or the paid features stay off and you buy only the analytics.
What you would look atA pilot agreement naming the instrument your school chose, the four things we record while it runs, the seat count and who pays for it in writing before day one, and the clause that says the result is published either way. Ask us on that first call for our contracting entity, W-9, certificate of insurance and cyber-liability position. Nobody outside our own team has verified those in writing yet; if we cannot produce them, stop.
Chief Academic Officer
Concept-level diagnosis, from work students do inside PrepGraph and nowhere else.
Every night PrepGraph places each activated student — an account that has been created and used, not a name on a roster — into one of three bands on every concept it has evidence for, ranks that class's concepts by which to reteach first, and names the students behind each one. Three limits you should hear from us rather than find in month three. Our library was built for Indian school and entrance-exam curricula; we have ingested no US state's official standards and no question in our bank carries a standards code. There is no progress-over-time view — a nightly snapshot, plus a saved student set you can re-check against it. And marks your teachers enter from a paper test are stored and audited but carry no concept tag and never reach the model, so a school that does not drive student usage sees an empty grid.
What you would look atOne class on screen, its concepts ranked by the weighted rule above, counts of how many understand / are developing / are struggling under each, the names behind the struggling group, and the date the rollup was computed. The concept is named in our own taxonomy, not in your state's code. A class with no rollup reads as unmeasured.
Principal
A signal, a prefilled screen, and no second system to maintain.
From a flagged concept or student, PrepGraph takes the teacher to the assignment, the class or the student with the fields already filled. Nothing writes back: the control that looks like it would — labelled Intervene — deep-links and records nothing, deliberately. That means no second intervention-tracking workflow, and it also means PrepGraph holds no record that a teacher acted, so no analysis of ours can connect a reteach to a change in a band. If you need that link it has to come from your own MTSS records. We have not measured teacher time saved: there is no timer in the product and no survey that asks, so any hours-per-week figure from us would be invented.
What you would look atThree students flagged this morning, each carrying its reason: a misconception where the teaching strategies already tried have failed, five or more days without recorded activity, or an accuracy regression on a concept. One more thing to weigh before you put this in front of children: the tutor is a live voice session, we record it, and we searched our own codebase for a self-harm, crisis or safeguarding escalation path and found none.
Teacher
Before first period: who is stuck, on what, and what to reteach first.
PrepGraph does not replace you and it cannot mark your paper tests. It identifies a misconception from the specific wrong answer a student picked — a distractor we tagged in advance — rather than from a score, and it says nothing at all about a student who has not used it. Where a class is on paid seats, the voice tutor keeps working that concept with a student when you cannot get to them one to one. In a school deployment that has to mean seats bought by the school for every student in the cohort: we will not run a class where support is allocated by whose family paid.
What you would look atYou save the students behind one concept as a named set, and the set records the band each was on the moment you cut it. Two weeks later you reopen it and see, student by student, where each sits now on that concept. That tells you the band moved. It does not tell you your reteach moved it, and we will not write it up as though it did.
The five questions
What we can't prove, and where your students' data actually goes
We have no evidence that PrepGraph works. No study has begun, no result exists, there are no case studies, and nothing here is offered as evidence of effect. What follows is the product as it is. Every number carries a bracket saying what it was counted from and when: [code, 1 Sep 2026] means we read it out of our source, migrations or configuration on that date; [dev content bank, 29 Aug 2026] means it was counted in a development copy of our content database — not production, which we could not query. Treat those counts as as-built rather than as-deployed, and ask us to re-run any of them against production in front of you.
- 01
1. What learning problem does it solve?
A narrow one. PrepGraph does not teach your class; it answers the question a teacher asks on Monday morning. Overnight it bands each activated student on every concept it has evidence for, ranks the class's concepts by which to reteach first, and names the students behind each. A leader gets chapter-level mastery across every class in one table and a single at-risk register merging every class's flagged students, worst first. Misconceptions come from the specific pre-tagged distractor a student chose, not from a raw score. The problem it does not solve is a student who has not used it: every signal comes from tutoring, practice and platform-graded work inside PrepGraph, and teacher-entered marks never reach the model.
- 02
2. When should it be used?
After a unit is taught, as weekly triage, by the teacher who already teaches those students. It should not be used as a placement instrument, a screening tool, an eligibility determination, or as evidence in an IEP, a 504 plan or an MTSS tier decision — the band is a three-level cut on an ability estimate, not a validated score, and none of those uses has been studied. We are putting that warning inside the product on the at-risk register itself, because a register ranked worst-first is exactly what a school reaches for when building a tier list. One consequence to plan for: once your staff maintain these bands and flags, they are part of the education record you hold, and a parent may ask to inspect them and to challenge what they say. We have no parent-facing view and no amendment path, so until we build one your staff carry those requests by hand.
- 03
3. For whom?
High school first, and math and science before anything else. Our strongest US assessment content is college-admissions testing — 3,949 questions built from past ACT and SAT papers [dev content bank, 29 Aug 2026] — and our strongest US course content is a concept map covering 39 AP courses [same source and date]. We would decline an elementary pilot and you should want us to: our US K-12 practice questions are generated by our models, are not teacher-reviewed, and about 8% have passed a solvability check. Three further limits on who this reaches. It is a paid per-student product. We have commissioned no accessibility audit and have no VPAT, so for a student using a screen reader or keyboard-only navigation we cannot tell you it works. And US tutoring is English-only by policy — the tutor declines a request to switch languages — which is your exposure under Title VI and the EEOA, not just our product gap. A grades 6–8 pilot also puts eleven- and twelve-year-olds on the product, squarely inside COPPA, and we have no verifiable parental-consent mechanism.
- 04
4. For how long?
We do not know, and no one should tell you they do: there is no research, ours or anyone's, establishing a dosage for this product. What we can tell you is the mechanism, because it is a setting. A concept shows a band as soon as there is any graded evidence for it — there is no minimum response count in our code [code, 1 Sep 2026] — which means an early band can rest on one or two answers. Treat a fresh grid as thin, and ask us for the response count behind any cell before acting on it. What we would propose is a semester if your calendar allows, with the measurement set up before the first login, a mid-point look at activation and weekly active practice together, and a standing agreement that if the usage is not there we stop rather than let it run empty and produce a number that means nothing.
- 05
5. What evidence shows it works?
None. What we offer instead is instrumentation you own: four measures are implemented in our pilot tooling — activation, weekly active practice, completion, and score change on an instrument the school sets, matched student by student. Four, not six; two more were designed and never built. You choose the baseline instrument, we do not score it, and we will say the limit in the same paragraph as any result: a baseline-versus-final comparison cannot separate our product from ordinary teaching, from the school year, from test timing, or from the attention a pilot attracts. Under the ESSA evidence tiers your authorizer and your Title funding actually use, that is Tier IV at best — a rationale — and a pilot as designed would not reach Tier III. Any figure we publish arrives with how many students took part, how many were approached and declined, how many left, who paid for the work and who conducted it — and the honest answer to the last is that the write-up is vendor-authored: we have appointed no independent reviewer, and a council we fund cannot serve as one. We publish a null on the same terms as a positive result.
The pilot
The pilot: an instrument you own
A pilot with PrepGraph is an instrument, not a free trial. Its purpose is to produce a written result you can hand to your board, including the result where nothing moved. Say the status of the machinery plainly: our pilot tooling is written and tested, its production migration is recorded in our own ledger with an unknown applied status, and it has never carried a pilot to a published result for any school anywhere. Yours would be the first end-to-end run, and where it breaks that is our failure to report to you.
What failure looks like
Five ways this fails, named in advance so nobody argues about the definition afterwards. One: fewer than seven in ten students complete the baseline, so the pilot never starts — a failure of our setup and your calendar, far better caught before the measurement than after. Two: students are provisioned and do not log in — an activation failure, and usually ours. Three: students log in and do not practice — the most common death, and the one thing we cannot fix for you from a distance. Four: teachers see the queue every morning and nothing changes in the schedule — a failure of the intervention half, and worth knowing about your own building. Five: everything runs, students practice, teachers act, and the score does not move. That is a real finding about our software, and it gets published on the same terms as a good one.
Current readiness
What we cannot do, cannot prove, or have not built
Verified against our own source code, not against a policy document. If you are evaluating us for procurement, this section and the one after it are the two that matter.
Our own copy
Start with our own pages, because you would find them anyway. As of 1 September 2026 our product site, partners.prepgraph.com, still tells US visitors that PrepGraph produced "+41 pts average assessment score improvement per student, measured over 3 months", a 68% drop in mid-year disengagement, 11 hours of teacher time saved per week and 94% family satisfaction. None of that was measured. They are four hardcoded strings a marketing commit added in March 2026, with no dataset, query or cohort behind them, and the same page carries testimonials whose text reads "[Insert teacher testimonial about intervention signals and time savings]" attributed to "[Teacher Name]". Those two we confirmed rendering on the live page on 1 September 2026, and you can check them yourself. Read in our own source on the same date, and therefore true of the code even where a reader cannot see the string on a page: a FERPA answer that says student data is handled consistent with FERPA requirements, when we have no FERPA program and the word appears in exactly one file in our entire codebase — that marketing page; a 14-day trial our code does not grant, since it grants 7; a claim that our content maps to your state standards, when we have ingested none; six pilot metrics of which four are built; and a "Teaching Agent Mesh" of six specialised agents, which does not exist. Taking all of it down is the condition for this page going up. If you find any of it still standing — on our site, in a cache, in a deck, in an email from us — that copy is wrong, it is ours, and we want to know.
In one place
The complete list
Written down so it can be held against us, and so you do not have to assemble it from fifteen pages.
- 01 Nothing has been measured. No study has begun, no result exists, there are no case studies, and no figure on this site is offered as evidence of effect.
- 02 Our content library was built for Indian school and entrance-exam curricula. We have ingested no US state's official standards and no question in our bank carries a standards code. Our US K-12 practice questions are generated by our models, are not human-authored, are not teacher-reviewed, and about 8% have passed a solvability check.
- 03 Concept-level gaps come only from work students do inside PrepGraph. Marks a teacher enters from a paper test carry no concept tag and do not feed the model — a school that does not drive student usage sees an empty grid.
- 04 There is no progress-over-time view in the product. What ships is a nightly snapshot, plus a saved student set that can be re-checked against the current snapshot.
- 05 A school licence does not put the tutor in students' hands. The tutor, the mastery loop and document Q&A are paid per-student features; every student gets a 7-day trial, and on day eight an uncovered student is denied all of them.
- 06 A student appears in the analytics once their account is activated, not when their name is loaded onto the roster — and the activation path we ship today was built for Indian families.
- 07 Our production database and identity records are hosted in India. A US-resident deployment does not exist.
- 08 Student work is processed by third-party AI providers, listed by name and country on this page, including one that operates from China. We cannot yet tell you, in writing, whether any of them may retain or train on it.
- 09 Tutoring sessions are recorded. Audio, transcripts and uploaded photographs of student work are swept after 90 days; other records are deactivated rather than erased, a school cannot direct deletion of a named student's record, and there is no export or contract-end purge.
- 10 We have no FERPA program, no signed district data privacy agreement, no SOC 2 report and no VPAT, and no accessibility audit has been commissioned. None of the four prerequisites listed on this page has started.
- 11 We have no SIS, LMS, LTI or district SSO integration. Rostering is a staff-loaded file; sign-in is email and password, Google, Facebook or LinkedIn, with no domain restriction.
- 12 Our scope is a single school. There is no multi-campus, district or CMO rollup.
- 13 We have one named partner school, in India, and a live footprint measured in tens of students. No US school has used this, and we are pre-revenue in the United States.
- 14 PrepGraph convenes and funds the Council for Responsible AI and School Innovation. It has one member — its founding chair — no appointed ethics panel, no study under way, and it certifies nothing, including us. Any figure attributed to it before its first published study is fabricated.
- 15 Our research layer at prepgraph.ai is a prototype that has never been applied to a student, a teacher, a classroom or a syllabus. Nothing here rests on it.
- 16 Every number on this page carries a bracket saying what it was counted from and when. [code, 1 Sep 2026] means it was read from our source, migrations or configuration on that date. [dev content bank, 29 Aug 2026] means it was counted in a development copy of our content database, not production, which we could not query — treat those as as-built rather than as-deployed. A number without a bracket is our error; tell us and we will date it or delete it, and we will re-run any of them against production in front of you.
- 17 One access-control defect on a student learning endpoint was found during this review and its fix gates the launch of this page. We are auditing the remaining student-facing endpoints and will publish what that audit finds. Until it is finished we make no absolute claim of data isolation across every endpoint.
Disqualify us early
PrepGraph is not right for you if
Each of these is a limit in the product as it stands on 1 September 2026, not a negotiating position. If three of them are dealbreakers we are the wrong vendor this school year, and this page is the cheapest place for you to find that out. If one or two are, name them on the first call and we will tell you honestly whether they change before your decision date or whether they do not.
- You are an elementary school. Our bank has no exam-tagged questions for kindergarten through sixth grade, and K-5 practice items are model-generated at request time. A pilot built on those would fail in public, which is bad for you and worse for us.
- Your evaluation rubric has a line for standards alignment. We have ingested no US state's official standards and no question carries a standards code. We score zero on that line and will not argue about it.
- You are in Texas and you want TEKS. We hold no TEKS codes, and the row in our database carrying that name is scoped to Indian Class 11 and 12 chemistry and biology — and is the only US row our runtime will serve unguarded.
- You roster through Clever, ClassLink, OneRoster, PowerSchool, Infinite Campus, Google Classroom or LTI. We integrate with none of them. Sign-in is email and password or a consumer social login, with no district identity provider and no domain restriction.
- You need student data resident in the United States. Our production database and identity records are in India and a US deployment does not exist.
- Signature depends on a FERPA program, a signed student data privacy agreement, a SOC 2 report or a VPAT. We have none of the four, and none of them clears this school year.
- Your agreement requires the school to direct deletion of a named student's record, or to get your data back or destroyed when you leave. Deletion runs through the student's own account, most records are deactivated rather than erased, and there is no export or contract-end purge.
- You need written no-training and zero-retention terms with every AI provider before students touch the product. We cannot show you those today.
- You need a documented child-safety and crisis-escalation protocol for an AI that talks to children one to one. We searched our own codebase for self-harm, crisis and safeguarding handling and found none. Your staff are mandatory reporters and our product would be a disclosure channel with no defined path behind it.
- Your students need accommodations, or instruction in a language other than English. There is no IEP or 504 field, no extended-time setting, nothing that reads a service plan, and US tutoring is English-only by policy. If your population is heavily newcomer or heavily served under IDEA, this will not serve them and we would rather not take the money.
- You need parents to be able to see or challenge what the software says about their child. There is no parent-facing view and no amendment path.
- You expect your own tests to produce the concept gaps. Teacher-entered marks are stored and audited but carry no concept tag and never reach the model.
- You cannot commit to students actually practising inside the product. No usage, no data, empty grid, nothing to report.
- You need every student covered automatically by the school's licence after the pilot. A school-level grant does not switch on student access in our system, and there is no US school price yet.
- Your board expects a network-wide view. The top scope in our academic reporting is one school; a user account belongs to exactly one school; campus is not a dimension in any academic view; and putting several schools under one account to fake a combined view would let every administrator see every school's students. We will not set it up that way and you should not ask any vendor to.
- You want a vendor's outcome number for a board packet. We do not have one and we will not invent one.