Add automated verification to your AI. A mathematical foundation for guarantees and checks across study design, data preparation, statistical analysis, and claims.
Connect your AIVerified across the chain — under the stated assumptions.
Automated checks make the covered calculations and conditions explicit—so less has to be checked again by hand.
Trace each supported claim to its assumptions, mathematical basis and recorded execution.
Share the inputs, methods and verification records others need to inspect and reproduce the checks.
Each stage must preserve what the next stage needs. Composition theorems—results proving that guarantees from separate steps still hold when the steps are chained—connect these local guarantees to a supported final claim.
Preprocessing can change what a test means. The mathematical contract identifies which transformations preserve the properties needed for valid inference.
Carry numerical uncertainty through the calculation. Resolve a decision only when every value within the certified bound supports it.
Compare the observed result with the declared reference law: the range of results chance alone would produce under the study’s design. Calibration gives the evidence a mathematical meaning.
Testing many hypotheses needs a rule for the whole discovery list. Valid e-values (evidence scores that average at most 1 when there is no effect) and e-BH (the Benjamini–Hochberg procedure applied to them) provide an upper bound on the expected false-discovery proportion.
18,473
exclusion cases, every one checked
In a two-group count model, removing any set of up to three of 48 observations leaves the direction of the estimated effect unchanged—proved for every case and confirmed by a separate checker.
15,285
hypotheses at genome scale
Certified discoveries across 15,285 hypotheses in semi-synthetic genome-scale count data. Exact reference laws for every gene compute in 9 seconds for a 12 + 12 design.
+55%
true discoveries at the same error bound
Protected scoring raised average true discoveries from 4.08 to 6.34 across 99 held-out conditions, and the chance of at least one true discovery rose too—with the false discovery rate within the declared bound (common-U e-BH).
162,323
DESeq2 values rigorously enclosed
Normalisation and initial dispersion values from actual DESeq2 1.42.0 runs on public RNA-seq data, enclosed in rigorous bounds across five analyses.
10 / 10
known PFOA-responsive genes recovered
On public rat liver toxicology data, the verification rule recovered all ten known PFOA-responsive genes at 2.5 mg/kg and above.
98 / 98
verifier tests passed
An independently implemented verifier recomputes the covered checks with exact rational arithmetic and passes every test in its suite.
Exclusions that lead to the same analysis are merged: 697 states become 487 and the certificate shrinks by 29.6%.
For fixed discovery tables, the exact best achievable result is derived—and reached.
Necessary and sufficient conditions show when a simpler ordered score can replace a four-tier score without losing discoveries.
Centring and scaling learned from the data and applied to both groups leave the test and its guarantee exactly intact.
Interval arithmetic certifies entire regions of continuous parameters, not just sampled values.
From the design alone, compute the minimum number of discoveries certification can reach—before collecting a sample.
