Deep dive — methodology
Independent-verifier declaration — why the control plane cannot reside within the entity being regulated
The signed statement at the end of every $499 Snapshot. Not a checkbox — a personal, standing declaration by Kevin Luddy as principal of contrarianAI LLC. Here is what it says and why the specific words matter in a federal AI context.
The declaration verbatim
"I, Kevin Luddy, as principal of contrarianAI LLC, personally attest that:
1. contrarianAI is not an employee, contractor, equity holder, or vendor of the audited program office / prime contractor or any of its predictive-maintenance AI vendors (Palantir, C3 AI, Shield AI, Anduril, Booz Allen, Leidos, SAIC, CACI, or any similar).
2. Assessment tools (distributional-shape sensor, retrieval-auditor, tool-call-grader, prompt-drift-sensor, predictor-corrector, router-drift-sensor) use model families and evaluation methods distinct from those in production at the audited program office.
3. This Snapshot is based on the data-sample submitted by the buyer. Changes to the audited program office's infrastructure after the Snapshot date are outside this Snapshot's scope.
4. Underlying audit evidence is retained for the Snapshot horizon and will be made available for legitimate regulatory / audit / oversight inquiry with appropriate legal process.
5. This SAMPLE Snapshot uses entirely synthetic data. No classified, CUI, ITAR-controlled, or otherwise export-restricted information was used, retained, or transmitted. Actual client engagements are scoped to buyer-submitted data only; contrarianAI infrastructure is CMMC 2.0 Level 2 aligned with Level 3 path available on engagement scope requiring it."
Sample declaration — signed Kevin Luddy, Principal, contrarianAI LLC, 2026-07-22
The five clauses, unpacked
- No entanglement. Not an employee, contractor, equity holder, or vendor of the program office / prime or any DoD AI-sustainment vendor. The independence is structural — not a policy claim.
- Distinct methodology. Different model family, different math, different retention pipeline. The verifier is not merely re-running the sustainment vendor's own instrumentation.
- Bounded scope. Data submitted at Snapshot time. Post-Snapshot changes are not covered. Honest scope prevents false-positive comfort in an ATO package.
- Retention + production stance. Underlying evidence retained; available for legitimate inquiry with appropriate legal process. Not an offer to hand over records on demand — a statement of the discipline.
- Data-handling statement. Explicit no-classified / no-CUI / no-ITAR statement for the SAMPLE. Client engagements scoped to buyer-submitted data; CMMC 2.0 Level 2 aligned. Load-bearing for federal-vertical buyers.
Why "personally attest" matters in a federal context
Kevin Luddy's name is on the signature. Not "contrarianAI LLC." Not "the audit team." A named principal, standing behind the statement. That is how declarations that survive DoD IG interview, GAO auditor questioning, or a JAG deposition are structured — a named person, addressable, cross-examinable.
Big-4 / Booz-Allen / Leidos / SAIC / CACI AI-assurance engagements do this at $400K-$1.2M+ per engagement. The $499 Snapshot does the same discipline at 5-10% of the price by scoping tightly to one AI surface + producing one determination + retaining evidence.
What "control plane" means for federal AI
A control plane in systems engineering is the layer that observes and governs another layer's behavior, from outside that layer. Applied to a defense AI program:
- The AI-sustainment vendor's own monitoring is inside the vendor's plane. Vendor cannot credibly grade its own homework.
- The prime contractor's IT monitoring is inside the prime's plane. Same principle.
- The program office's own sustainment-analytics platform is inside the program's plane. Also the same principle.
- An independent verifier — distinct entity, distinct methodology, distinct retention — is outside all three. That is the control plane.
"We use our sustainment vendor's continuous-monitoring dashboard" is the modal answer GAO / DoD IG reject as insufficient for RMF continuous-monitoring under NIST SP 800-53 rev5. The Snapshot exists so that a defensible answer — independent verifier, signed declaration, retained evidence — can be produced at $499.
$499 Snapshot. 3 business days.
Signed independent-verifier declaration on your program office's actual AI surface. Retained evidence. Named principal. CMMC 2.0 Level 2 aligned.
Buy $499