Vertical Intelligence is the queryable evidence layer for every public school district in the country: what it teaches, whether its students actually improved, the budget a switch is fundable from, the accountability pressure it's under, when its buying window opens, and who signs. Every dataset is joined by the same district key and computed from government data the vendor can't edit. Ask it in plain English from any AI assistant, and every number comes back traceable to its public source. Watch it run the questions a curriculum provider, a district, a funder, and an investor actually ask.
Try asking
Illustrative demo · Ohio incumbents, California & E-Rate funding, fiscal distress, and the Maryland checkbook are live PILLAR data; estimates are intervals, never verdicts. Hover to pause.
The question the market can't answer
A Math or ELA platform wants to unseat an incumbent. The one fact that would win the room, whether student outcomes actually moved under the program already in the building and who can fund a switch, sits scattered across dozens of public sources, never computed, and never in the room when the decision gets made. So the market rewards the familiar logo. Vertical Intelligence pulls that evidence together and serves it into the decision.
District adoptions, independent government outcomes, and the controls for a fair comparison, all linked by district.
Difference-in-differences with a parallel-trends pre-test. Every result an interval with its confound flag, backtested out-of-sample.
The whole layer exposed as plain-English tools. No SQL, no dashboard to learn, no intake form to lobby.
Any AI assistant, whether Claude, ChatGPT, or an internal copilot, queries VI right where the decision gets made.
What Vertical Intelligence knows
Vertical Intelligence connects what each district teaches, how its students are doing, the budget it has to spend, and the rules it has to follow. Everything is linked district by district, so a single plain-English question can draw on all of it at once.
SEDA plus a NAEP-anchored owned layer (r=0.94). A national outcome ruler the vendor can't edit.
District-level curriculum choices, parsed to publisher and edition. Real adoptions, not a state menu.
Title I-A, state categoricals, and per-school LCFF S&C, SENI, and ESSA spend. The budget a switch is fundable from.
ESSA designations and proficiency across 51 jurisdictions. The regulatory pressure and the academic gap.
SAIPE poverty, enrollment, FRPL, and neighborhood income. The controls that make a comparison fair.
Fiscal-year calendars, budget-approval months, PO blackouts, and the RFP window per state.
Need, ability to pay, pressure, and timing, weighted for durability. Who is actionable now.
The people behind the district, pulled from official directories, so the signal reaches a human.
51 jurisdictions · every dataset linked by the NCES district identifier · governed by a Spec → Guarantee → Test framework so the method can't be quietly re-weighted.
Who asks Vertical Intelligence
Lead a switch with evidence, not a deck, and walk in with the district's own fundable budget.
"Where has the incumbent failed to move outcomes, and who can pay to replace it?"
Measure every program, including your own portfolio, by a method you don't control.
"Which interventions actually move outcomes for the students we fund?"
Bring decision-point evidence computed from your peers' real outcomes to the adoption table.
"Did this program work for districts like mine, under our conditions?"
Independently verify an EdTech efficacy claim against government outcome data.
"Does the efficacy story hold up on data the company can't edit?"
Why it holds up
The outcome layer is one independent, nationally comparable measure of student results for every public district in the country, computed from government data no vendor can edit, by a method written as test-enforced invariants so the rubric can't be quietly re-weighted. Funding and procurement are national too. On top of that national base, two deeper layers go as far as each state's own records allow, and the list keeps growing: the named program in each district, and the actual checkbook.
Independent outcomes, funding, and procurement cover every district in the country. Where a state publishes more, VI goes deeper: twenty states and Washington DC name the reading or math program districts teach (statewide in most, the largest districts in the rest, more than 3,100 districts with a named incumbent), ten states show who runs high-dosage tutoring, and four open the full district checkbook, the real payments read across more than a decade. New states come online as their records allow.
The Legibility Series · two essays
Why the market can't reward what it can't see, and the infrastructure that changes that. A response to AEI's The $30 Billion Question, with the proof of existence behind this demo.
One state's checkbook, read across sixteen years: more than a quarter of a billion dollars spent on teaching and learning, and no measurable movement in outcomes.
See Vertical Intelligence run your category, your states, and your incumbents, in plain English on independent data.