The exact question, response, verdict, and failure pattern—not just an aggregate score.
Independent AI audit evidence · on your infrastructure
When your AI is questioned, bring the record.
VeriTrooper examines source documents and structured data, tests AI answers against the material you approve, and preserves findings for human review. Run with local models, inspect deterministic checks, and verify signed evidence packages offline.
ILLUSTRATIVE RECORDVT / AI-AUDIT / 042
EVIDENCE FILEOne verdict.
Every reason attached.
Every reason attached.
- Audit basis
- Approved source material
- Test article
- Real model interaction
- Finding
- Answer · verdict · citation
- Verification
- Portable offline package
REVIEW STATEPASS / FAIL / WHYRecorded scope · inspectable checks
Source-data inspectionDeterministic checksLocal operationOffline evidence verification
THE VERITROOPER AUDIT SYSTEM
From approved source to reviewable audit record.
Examine the sources. Test the answers. Inspect the findings. Preserve the record.
AUDIT BASISApproved source materialDocuments · policies · records · expected behavior
TEST · TRACE · REVIEWAuditable findingsConnect each verdict to the tested interaction, supporting evidence, scope, and any recorded human review
REVIEWABLE OUTPUTVerifiable evidence packageFindings · sources · scope · review record · offline integrity verificationInspect the evidence →
What the audit gives you
A record you can act on—and defend.
The approved source, rule, or expectation used to evaluate the interaction.
Evidence that helps the team locate problems in prompts, retrieval, policy, source data, or model behavior.
A repeatable run record that shows what changed when the corrected system is tested again.
One assurance system · three checkpoints
Test the data. Audit the answers. Monitor what ships.
Before you build
SitRep
Inspect documents for conflicting claims and structured data for inconsistent values, missing cells, and missing keys.
Governance Records workflow includedExplore SitRep →
Before you ship
Scout
Test calculations, precision, conditions, exceptions, and refusal behavior against source-linked references.
Governance Records workflow includedExplore Scout →
After you ship
Watchtower
Review samples of captured conversations or generated endpoint probes on a schedule, preserving findings and recurring patterns.
Governance Records workflow includedExplore Watchtower →Correct with precision
Move from “the score is low” to “this is what needs correction.”
VeriTrooper ties findings to the tested interaction, supporting source material, and recorded checks. Use that evidence to investigate source data, retrieval, prompts, or model behavior, then compare results after a change.
See how findings are produced →01
Locate the failure
Preserve the question, answer, verdict, and observed failure
02
Trace it to evidence
Identify the source or expectation the answer did not satisfy
03
Correct the system
Use the record to target prompts, retrieval, policy, data, or model behavior
04
Retest the change
Rerun the audit and preserve the new result beside the original
EU AI Act evidence
Turn test results into review-ready technical documentation.
VeriTrooper organizes measured accuracy, robustness, limitations, testing procedures, dated run records, and human signoff into evidence that supports Article 11 and Annex IV documentation.
- Annex IV testing and validation record
- System card, data card, and framework crosswalk
- Traceable findings, run configuration, and reviewer disposition
Governance Records workflow · included with all three products
Carry each product's evidence through accountable review.
The shared Governance Records workflow combines audit evidence with system facts, human classification decisions, oversight measures, owners, and remediation. It is a supporting workflow, not a fourth product. Your reviewers supply the decisions; guided professional services are quoted separately.
- Verified evidence-package intake from any core product
- Recorded classification, exception, and remediation decisions
- Named owner, reviewer, approver, legal entity, and review date
Recorded baseline and pipeline results
Measured comparisons, preserved as evidence.
These results compare the recorded configurations on the same eligible questions. Inspect the run conditions and findings before applying them to another system.
88.08%SEC 10-K · gpt-5.6-solfrom 72.89% · n=889 · 26 Aug 2026
96.60%IRS tax code · claude-fable-5from 88.40% · n=1,000 · 26 Aug 2026
96.60%FDA drug labels · gemini-3.1-pro-previewfrom 90.60% · n=1,000 · 26 Aug 2026
89.96%Combined corpus · local air-gapfrom 84.74% · n=249 · 15 Aug 2026
Evaluation history · April to September 2026
Documented repeatability. Traceable results.
VeriTrooper maintains months of evaluation history, including two ten-run repeatability studies and preserved records of failures, corrections, and retesting. Reviewers can trace reported results back to their supporting evidence.
236 / 238question verdicts stable across all ten July UCMJ runsAdopted scoring basis, under the recorded test conditions.
Why the audit holds up
Inspect how each answer was judged.
Scout uses deterministic checks for arithmetic, units, precision, and other supported rules against stored references. Diagnostic opinions, optional cross-vendor review, and referrals remain visible alongside the headline result. A qualified reviewer can record signoff after examining the findings.
Read the assurance method →01
Source-linked references
Review the questions, expected answers, and supporting material
02
Deterministic checks
Reproducible rules settle clear cases
03
Optional model review
The record identifies the models used and any shared roles
04
Human signoff
A named reviewer can record a decision after review
CUSTOMER ENVIRONMENTSourcesVeriTrooperEvidence package
↔
OPTIONALCustomer-selected model endpoint
Your environment · your evidence
Run the audit where your data lives.
Use the bundled local model and deterministic checks for local operation. Retain findings on your own storage and verify package integrity offline. Optional remote model roles send the required material directly to the providers you select.
Review security and deployment →Professional · $499.99 USD · one-time
Your audit suite. Your evidence.
Scout, SitRep, Watchtower, and Shared Governance Records for individual professionals. Local processing, evidence exports, and basic technical support. Company deployment and guided assessments are quoted separately.
Windows · macOS · Linux Evaluation Suite
Try all three audit products.
14 days from the first real audit, five shared runs, and up to 200 questions or evaluated interactions per run. Request the free trial and receive the installer for your platform by email.
10-business-day guided assessment
Bring one important AI workflow.
One product, one use case, one approved source corpus, and up to 250 questions or monitored turns, ending in a verifiable package and executive readout.