Amendment 002 · Filed 2026-04-16
Amendment No. 002 to the Operating Agreement of The Cochran Block, LLC
Revenue Model Articulation, Business Acceleration Framework,
and Mission-Aligned Earnings Authorization
Codename: More Money Less CSAM (MMLC)
Effective Date: April 16, 2026 · Companion to Amendment 001 (CEBP)

Section I · Recitals

WHEREAS Amendment 001 authorized the Chicken and Egg Bypass Protocol (CEBP), permitting one-time, capped, sunset-bound acceptance of dilutive funding via the a16z Speedrun accelerator;

WHEREAS the CEBP sunset conditions in Amendment 001 §12.4(d) include achievement of Five Million Dollars ($5,000,000) in trailing-twelve-month gross revenue, requiring the Company to articulate and authorize the revenue streams that will produce such revenue;

WHEREAS the operational mission of KNOXAI — eliminating contaminated AI models from the public ecosystem, with CSAM as the headline harm class and ten additional classes covered per cochranblock.org/no-quarter — produces measurable harm reduction in direct proportion to operational scale, which scales with revenue;

WHEREAS the Member determines that more revenue equals more operators on payroll, equals more models audited, equals more publishers reported, equals less CSAM in the public AI ecosystem — and accordingly that pursuit of revenue across multiple authorized streams is mission-aligned, not mission-divergent;

WHEREAS the codename "More Money Less CSAM" (MMLC) is selected to make explicit the tight coupling between business growth and harm reduction, and to communicate the framework to investors, operators, regulators, and the public in language that resists corporate softening;

WHEREAS this Amendment further establishes Protocol P32 (Mission-Aligned Revenue) to govern revenue acceptance and ensure no revenue stream is pursued that would compromise the integrity of the audit methodology, the operator guild, or the reporting pipeline; and

WHEREAS this Amendment is executed pursuant to Article XV of the Original Agreement.

NOW THEREFORE, the Original Agreement is hereby amended as follows:

Section II · Revenue Stream Authorization

A new Section 12.5 is hereby added to Article XII, immediately following Section 12.4 (CEBP):

12.5 Authorized Revenue Streams

The Company is hereby authorized to pursue and accept revenue from each of the following enumerated streams, all of which are deemed within the scope of Article III (Purpose, Powers, and Flexibility) of the Original Agreement:

(a)Cert Fees. Tiered audit certification fees per the published pricing schedule (Standard, Operator, Full Spectrum, Portfolio, Gov), including per-class add-on fees.
(b)Hash Database API Licensing. Subscription access to the curated bad-content hash database for third-party platforms, T&S teams, and adjacent audit services.
(c)Continuous Monitoring Subscriptions. Recurring monthly re-audit programs for Portfolio and Enterprise customers throughout cert validity windows.
(d)Insurance Partnership Revenue. Revenue-share arrangements with E&O / cyber-insurance carriers offering premium discounts for KNOXAI-certified AI products.
(e)Government Contracts. Federal, state, and allied-government contracts including but not limited to DoD SBIR/STTR/OTA, DHS S&T, IC procurement, and DARPA programs.
(f)Training and Certification Programs. SkillBridge cohort sponsorships (DoD-funded), VR&E partnerships (VA-funded), university curriculum licensing, and commercial bootcamp tuition.
(g)Expert Witness and Litigation Support. Operator-billed forensic AI testimony, deposition preparation, and litigation-support engagements for civil and criminal proceedings.
(h)Regulatory Consulting. Engagements with enterprises building internal AI integrity programs, government agencies developing AI safety frameworks, and standards bodies drafting conformity protocols.
(i)Marketplace Integration Fees. Bulk-cert pre-purchase, integration engineering, and listing-gate revenue from AI model marketplaces (HuggingFace, Civitai, Ollama, Replicate, Together AI, and successors).
(j)Dataset Audit Services. Pre-publication audit of public training datasets for the dataset publisher or for downstream consumers.
(k)AIBOM Standard Licensing. Licensing of the AI Bill of Materials format and validation tooling to enterprise procurement teams, regulators, and auditors.
(l)Foundation and Charitable Grants. Mission-aligned grants from private foundations, charitable trusts, and equivalent non-dilutive vehicles.
(m)Honoraria and Speaking Engagements. Speaking fees for engagements consistent with Section 21.5 (Persona-Not-A-Product clause) — i.e., fees paid for the Member's expertise applied to a specific question, not for the Member's persona as a brand asset.

The enumeration above is illustrative of authorized streams under Article III; it is not exhaustive. Future revenue streams not listed are authorized provided they (i) pass Protocol P32 (Mission-Aligned Revenue, established below), and (ii) are reflected in the Company's books pursuant to Article VIII.

Section III · Addition of Protocol P32 (Mission-Aligned Revenue)

Article XVIII of the Original Agreement is amended to include the following new Protocol, inserted after Protocol P31 (Unbreaking Software):

P32 — Mission-Aligned Revenue

Every proposed revenue stream shall be evaluated against the following question:

"Does this dollar increase, decrease, or leave unchanged the platform's ability to find and stop contaminated AI models?"

Revenue streams that increase that ability are authorized and pursued. Streams that leave it unchanged are authorized only if they consume no operator attention. Streams that decrease it — including but not limited to consulting engagements that conflict with audit independence, customer payments contingent on a finding outcome, and revenue arrangements that compromise the operator guild's reputation — are NOT authorized.

The Member shall maintain a running record of revenue acceptance decisions, with the P32 evaluation noted for each material new stream. The record is internal to the Company but shall be made available to TAC members upon request.

Tag commits and contracts implementing P32 with [P32].

Section IV · Schedule A Incorporated

Schedule A (Business Model Annex — More Money Less CSAM Framework) is hereby incorporated into and made part of this Amendment. Schedule A includes market sizing, the pricing matrix, the five-year revenue projection, unit economics, defensibility analysis, allocation of CEBP capital under §12.4(h) of Amendment 001, and risk register with mitigations. Schedule A is intended to function additionally as the Company's investor-facing business annex for purposes of the a16z Speedrun submission authorized under CEBP.

Section V · Integration with Original Agreement

(a)Survival. Except as expressly amended, all other provisions of the Original Agreement and Amendment 001 remain in full force.
(b)Conflicts. In the event of any conflict between this Amendment and prior governance documents, this Amendment controls in matters of revenue stream authorization and Schedule A's interpretive framework.
(c)Sunset Coordination. Section 12.5 (Revenue Stream Authorization) is permanent and survives the CEBP Sunset defined in Amendment 001 §12.4(d). Revenue streams authorized herein may continue to be pursued post-Sunset under the Original Agreement's strict non-dilutive posture.
(d)Public Disclosure. This Amendment, including Schedule A, shall be publicly published at cochranblock.org/operations/amendment-002 within twenty-four (24) hours of execution.
(e)Governing Law. Maryland, per Article XVI of the Original Agreement.

Section VI · Execution

The undersigned Member, being the sole Member of The Cochran Block, LLC, acting in such capacity and pursuant to the authority granted in Article XV of the Original Agreement, hereby executes this Amendment as of the Effective Date set forth above.

/s/ Michael Cochran
Michael Cochran
Sole Member · The Cochran Block, LLC
Date: April 16, 2026
ACKNOWLEDGMENT OF PUBLIC DISCLOSURE
This Amendment, including Schedule A, is published publicly at cochranblock.org/operations/amendment-002
in compliance with Section V(d).
Schedule A · Business Model Annex
More Money Less CSAM
The KNOXAI Revenue Framework — every dollar earned funds an operator who finds another contaminated model

1 · Market Sizing (Company Estimates — See §12 Factual Basis)

~$10B
TAM Est. by 2027
~$2-3B
SAM Est. (Audit subset)
$40M
SOM Y5 Internal Target
~1.5%
Target SAM Capture Y5

TAM estimate: The Company's internal estimate of the 2027 global AI safety / model integrity / responsible AI services market, derived from publicly available industry analyst commentary. Specific cited third-party market-sizing reports are not relied upon here; the estimate is directional only. The market includes red-teaming services, model-card auditing, dataset provenance, model-monitoring tools, regulatory-compliance consulting, and training programs.

SAM estimate: The audit-and-attestation subset — independent third-party assessment of AI model integrity for regulatory, contractual, or marketplace requirements. KNOXAI competes here directly. Subset size is an internal allocation of the TAM estimate; not sourced from a third-party report.

SOM target: Internal target for Y5 revenue capture, assuming (i) penetration into ~30% of HuggingFace's publicly stated ~1.5M+ public models at Standard tier, (ii) one Federal SBIR Phase II + one DHS contract, (iii) two insurance partnerships, (iv) one marketplace integration, (v) 30 Portfolio-tier customers. No assumption of legislation-mandated adoption. These are projections, not commitments.

2 · Pricing Matrix (per §12.5(a))

TierAnnual PriceWhat's IncludedOperator Split
Standard$20Automated CSAM hash + generic Gate 5 · any operator signs80% / 20%
+ NCII Add-on+$15Identifiable-person deepfake audit75% / 25%
+ CBRN Add-on+$30Bioweapon / chemical / radiological battery (biosec specialty)70% / 30%
+ Supply-Chain+$25Backdoor weight sweep · membership inference rigor70% / 30%
+ Privacy+$15PII memorization · GDPR/CCPA exposure75% / 25%
+ Copyright+$20Training-corpus provenance · DMCA exposure75% / 25%
Full Spectrum$150All eleven harm classes bundled70% / 30%
Operator$500Customer picks a specific named operator70% / 30%
Portfolio$5K - 50KDedicated operator(s) · full model shelf · monthly re-audit60% / 40%
Gov / Defenseper engagementClearance-required · multi-operator dual-signed certs50% / 50%

3 · Five-Year Revenue Projection (Internal Forecast — Not a Commitment)

Disclosure: the table below is the Company's internal revenue forecast under stated assumptions. It is not a projection supported by booked contracts, historical customer data, or third-party market research. Actual results will differ, likely materially. The CEBP investor (a16z Speedrun) and any other recipient of this document should treat all forward-looking numbers as targets, not commitments.

StreamY1Y2Y3Y4Y5
Cert Fees (Standard + add-ons)$50K$300K$1.2M$4.5M$12M
Operator Tier$15K$80K$250K$700K$1.5M
Full Spectrum$10K$60K$240K$900K$2.4M
Portfolio Subscriptions$50K$250K$800K$2.5M$6M
Gov / DoD Contracts$300K$1.5M$3M$8M
Hash DB API Licensing$50K$300K$800K$2M
Insurance Partnership$150K$700K$2.5M
Training (SkillBridge / Univ / Comm)$100K$300K$700K$1.5M
Expert Witness / Forensic$30K$120K$300K$700K
Marketplace Integration Fees$100K$500K$1.5M
Foundation Grants (non-dilutive)$100K$400K$500K$700K$1M
Dataset Audit + AIBOM Licensing$50K$300K$1M
Total Annual Revenue$225K$1.57M$5.51M$15.6M$40.1M
Year 3 trips CEBP Sunset trigger (i): $5M ARR. Strict non-dilutive posture resumes at end of Year 3.

Notes: Revenue projection assumes (i) Speedrun close in Q3 2026, (ii) SBIR Phase I award in Q1 2027, Phase II in Q3 2027, (iii) one significant marketplace integration (HuggingFace or Civitai) by end of Y2, (iv) one insurance partnership operational by Y3, (v) NO state or federal AI-CSAM legislation citing KNOXAI by name — that would be upside not modeled here.

4 · Unit Economics

MetricStandardOperatorPortfolioGov
Avg Annual Contract Value (ACV)$25$500$15,000$250,000
Customer Acquisition Cost (CAC)$3$25$1,500$25,000
Operator Compensation per Cert$20$350$9,000$125,000
Platform Take per Cert$5$150$6,000$125,000
Gross Margin~85%~80%~75%~65%
LTV (3-year)$60$1,300$40,000$650,000
LTV / CAC20x52x27x26x

Standard tier CAC is low because acquisition is viral via marketplace integrations and GitHub Action distribution. Operator tier CAC is low because customers self-select operators from the public directory. Portfolio and Gov tiers carry sales cycles measured in months and matched CAC, but ACV justifies it.

5 · Defensibility & Moat

Moat ComponentSource
Hardware-rooted attestationSoftware-only competitors cannot replicate eFuse-bound signing keys; the cert architecture is structurally distinct.
NCMEC partnership (post-Y1)Regulatory moat. Single federal partnership creates de-facto standard. Peer competitors face years-long approval queue.
Operator network effectsMore operators → broader specialty coverage → better cert quality → premium pricing → more operators recruited.
Federal legislation citationNaming KNOXAI in state or federal AI-CSAM legislation creates statutory moat. Active outreach in Y2-Y3.
Veteran-owned, OCO-vetted brandFounder credibility (USCYBERCOM, 100+ missions) is non-replicable by VC-funded software-engineer founders.
Operating Agreement transparencyPublic OA with binding governance differentiates from typical compliance vendors. CEBP & MMLC frameworks publicly disclosed builds deeper trust.
Reproducible audit artifactsEvery cert is independently verifiable by skeptics. Trust does not require trusting the platform.

6 · Why a16z Speedrun Specifically

The Member is not raising for capital alone. Per CEBP §12.4(b), the dilution cap is 12%. Speedrun's $750K offer represents capital to bridge from founding to first non-dilutive replacement (SBIR Phase II in ~9-12 months). What Speedrun specifically supplies that other capital sources cannot:

  1. a16z portfolio access. a16z's American Dynamism, AI, and infrastructure portfolio is the highest-density set of likely Portfolio-tier and Gov-tier KNOXAI customers in the United States. Speedrun cohort = warm intros to 50+ companies.
  2. Founder community velocity. Speedrun cohort timing accelerates operator-recruiting through peer founders' networks.
  3. Validation signal for non-dilutive raises. Knight Foundation, Mozilla, and Schmidt Futures fund a16z-backed AI safety startups at higher acceptance rates than cold applicants. Speedrun acceptance compounds the non-dilutive raise that follows.
  4. Capital tempo. Speedrun closes in weeks; SBIR Phase I in months; foundation grants in 3-6 months. Speedrun bridges the gap.
  5. American Dynamism thesis fit. Veteran-owned, national-security-adjacent, child-protection mission, hardware-rooted, sovereign-architecture. KNOXAI is exactly Katherine Boyle's published thesis instantiated.

7 · Use of CEBP Capital ($750K)

AllocationAmountPurpose
Founder fair-market salary (12 mo)$250KPer CEBP §12.4(h) — full-time founder compensation at fair-market rate to enable single-pointed focus
Operator FTE bench (2 × 6 mo)$200KTwo part-time operators at $100K annualized, bridging from volunteer guild to paid bench
SBIR Phase I prep + matching consultant$60KPhase I application + grant-writing support; replaced by Phase I award (~$200K) on success
NCMEC partnership application + legal$50KFederal partnership filing, attorney review, vetting period costs
Hardware infrastructure (vaults + signers)$50K10 operator hardware kits ($500 ea) + 5 reference vaults + signing infrastructure
Foundation grant applications$30KApplication fees + grant-writing for Knight, Mozilla, Ford, Hewlett, Schmidt Futures
Marketplace integration engineering$50KHuggingFace + Civitai integration sprints; opens marketplace channel revenue
Insurance partnership outreach$30KConferences (RIMS, AIA), introductions to Coalition / Vouch / Beazley underwriters
Legal counsel (cert defense + IP)$30KRetainer for cert-dispute defense, IP filings on AIBOM standard
Reserve / contingency$0Per discipline; if reserve needed, must come from earned revenue, not Speedrun draw
Total$750KSum of allocations

8 · Risk Register & Mitigations

RiskMitigation
Big AI labs build CSAM detection internallyInternal teams aren't independent third parties. Regulators, marketplaces, and insurers want external attestation. KNOXAI sells the independence.
Regulatory shift obviates the certActive legislative outreach (Lever 06 in /no-quarter doctrine). Become the regulator's preferred conformity path before any standard solidifies.
Operator quality drift / bad cert issuedTAC peer review, random re-audit program, expulsion mechanism with public delisting. Quality drift = expulsion = guild reputation preserved.
Mission drift toward revenue extractionP32 (Mission-Aligned Revenue) explicit safeguard. Every revenue stream evaluated against impact on detection capability.
Founder burnout / single point of failureOperator #1 onboarding accelerates redundancy. TAC structure designed for distributed leadership. OA Section 10.X (Indemnification) provides personal liability shield.
Speedrun dilution leads to follow-on equity pressureCEBP §12.4(c) explicitly prohibits follow-on. Sunset enforcement is structural, not aspirational. Reaching $5M ARR triggers automatic non-dilutive return.
Adversarial LARPing — bad actors register as operators to fail-cert competitorsTAC vetting before operator joins directory. Public reputation scoring. Random platform-issued canary certs to detect lazy or compromised operators.
Hardware key loss / theftDocumented key rotation procedure (CEBP-adjacent governance). Operator-level only — platform never holds keys, so platform-level catastrophe is impossible.

9 · The Money-vs-CSAM Conversion Rate

The MMLC framework's central claim: every dollar of revenue translates directly to operator-hours available for audit, which translate to contaminated models detected, which translate to regulator reports filed. The chain is:

  • Platform take → operator bench capacity
  • Operator bench → cert audits produced per cycle
  • Cert audits → contaminated models detected at the prevailing fail rate (to be measured in production)
  • Detections → mandatory-reporting filings to the authority with jurisdiction
  • Filings → investigation pipeline downstream of KNOXAI's control

Conversion rates at each stage are mission-dependent and will be measured once operational. The Company does not here cite investigation or indictment rates because authoritative public base rates for AI-CSAM specifically do not exist in 2026; the category is too new. The live ledger at cochranblock.org/no-quarter will publish measured rates post-launch as they become defensible.

What is factual: mandatory reporting is triggered under 18 USC §2258A on any CSAM finding. What KNOXAI measurably contributes is the upstream detection signal that would otherwise not exist — the regulator's conversion of that signal into enforcement is out of scope.

10 · Exit Considerations

Member's stated exit posture: The Cochran Block, LLC is a perpetual entity per Article IV §4.1 of the Original Agreement. The Member does not contemplate sale, acquisition, IPO, or dissolution. The CEBP-permitted Speedrun investor (a16z) acquires a minority stake under typical SAFE terms; the SAFE converts in a future priced round if and only if such a round occurs (and per Amendment 001 §12.4(c), no such follow-on is authorized without separate Article XV process).

Practically: a16z's exit on the Speedrun investment is realized through (a) secondary sale of the SAFE to another party at a future negotiated valuation, or (b) a buy-back by the Company funded from earned revenue at a negotiated price post-Sunset. Both paths are compatible with the Member retaining 100% control of the Company indefinitely.

This posture is unusual for a Speedrun investment but is disclosed transparently up front. Investors uncomfortable with this posture should not subscribe.

11 · The Close

The amount of CSAM in the public AI ecosystem
is inversely proportional to the revenue of this company.
More money. Less CSAM.

That's the framework. That's the codename. That's the math. Schedule A is the documentation that the Speedrun investment, the dilution cost, the founder's salary, and the operator-bench investment all serve a single optimization function — and that function is denominated in indictments filed and contaminated models stopped, not in EBITDA.

If that framing is compatible with the investor's thesis, the Company welcomes the partnership under the terms enumerated in Amendment 001 (CEBP) and the revenue authorization established in this Amendment 002. If it is not, no investment shall be accepted.

Original OA  ·  Amendment 001 (CEBP)  ·  Manifesto  ·  Doctrine