AI-drafted, expert-reviewed items mapped to the current digital SAT specification
Request a sample set or start a pilot
A four-stage production pipeline for test-preparation providers, publishers, and district assessment teams

Every item is drafted against a coverage table built from the published test specification, so each domain and skill in the digital SAT Suite has a slot — and unmapped items are rejected before they can ship.
Each item, its options, and its rationale are drafted by an AI pipeline from the source material used, with a citation attached to that source.
Machine checks confirm exactly one defensible answer, no overlapping options, a named skill mapping, and correct units, figures, and notation before any item advances.
No item ships without review by a subject-matter expert, who reads, edits, or rejects it. Machine validation checks form and logic; the reviewer judges whether the item is a fair measure of the skill — neither replaces the other.
Items ship in sets of 20, four options each, complete with a coverage table and rationales that justify the key and explain each distractor, in HTML, PDF, and a structured interchange file.
Items are drafted against the current test format used in the 2024–25 SAT Suite — covering the PSAT 8/9, PSAT 10, PSAT/NMSQT, and SAT — so practice material matches what students actually sit for.
| Product | Digital SAT Practice Item |
| Customers | Test-preparation providers, publishers, schools and district programs |
| Assessments covered | PSAT 8/9, PSAT 10, PSAT/NMSQT, SAT |
| Production stages | 4: specification mapping, drafting, deterministic validation, expert review |
| Specification mapping | Coverage table of domains and skills; every skill in the specification has a slot |
| Deterministic validation checks | Exactly one defensible answer; no two options mean the same thing; each item maps to a named specification skill; units, figures and notation follow style rules; rationale states why the key is correct and why each distractor is not |
| Expert review | Every item signed off by a subject-matter expert before release |
| Delivery unit | Sets of 20 items |
| Options per item | Four options each |
| Rationales | Key justification plus a reason for each distractor |
| Coverage table (delivery) | Item to specification skill mapping included with every set |
| Formats | HTML, PDF, and a structured interchange file |
| Order-time customization | Domain, skill and difficulty specified at order time |

How test-preparation providers, publishers, and district assessment teams put Qraft to work
Test-preparation providers, publishers and school or district programs turn to Qraft when they need practice items that match the current digital SAT Suite rather than the retired paper format. The 2024–25 school year was the first full school year of digital testing across the PSAT 8/9, PSAT 10, PSAT/NMSQT and SAT, and demand has not slowed: SAT participation in the class of 2025 surpassed 2 million test takers, the first time since 2020. Programs still relying on pre-2024 material are practicing students against a format that no longer exists, and replacing a bank item by item is slow.
For test-prep program directors, this means sourcing new items that reflect the current test specification without waiting on a manual rewrite of an entire bank. Each Qraft item is drafted from the published test specification, screened by deterministic checks for a single defensible key, non-overlapping options, correct skill mapping and correct style, and then reviewed by a subject-matter expert before release — so directors receive items that are already vetted for both form and pedagogical fairness.
For publishers, delivery is structured to slot into existing production workflows: items ship in sets of 20, four options each, with a coverage table mapping every item to a specification skill and a rationale that justifies the key and explains each distractor. Sets are available in HTML, PDF, and a structured interchange file, and domain, skill and difficulty can be specified at order time.
For district assessment leads, the same four-stage process — specification mapping, drafting, deterministic validation, and expert sign-off — gives a documented basis for confidence in item quality: no item ships without a human reviewer's sign-off, and machine validation and expert judgment are treated as complementary rather than interchangeable steps.
For test-preparation providers, publishers and district assessment leads
Pick a domain and take one sample set first. Twenty items arrive with the coverage table and a rationale for every key and distractor. Review the sample, align the review criteria with your editors, then agree a delivery schedule and volume.
Request a sample set or start a pilotSample requests and pilot enquiries for test-preparation providers, publishers and district programs
Request a sample set or pilot