Claims substantiation + pre-approved claims library for AI content
Per-operator approved-claims library with substantiation evidence — every AI content output checks each substantive claim before it ships.
The problem
Your VP-Marketing says "clinically proven." Your founder says "most effective." Your supplements page references a 2019 study legal has not vetted. Your finance copy says "best-in-class returns." Each of these is a claim. Legal needs to substantiate every one — clinical study, comparative testing, performance backtest, FDD Item 19 methodology — or it has to come down.
The FTC, FDA, and state AGs enforce against unsubstantiated claims. Class-action plaintiff lawyers file lawsuits over them. Your AI content agents generate copy with claims nobody on legal ever approved. Even your human writers do this — they grab phrasing from last quarter's campaign without knowing whether the underlying substantiation is still valid.
Westlaw and LexisNexis at $5,000-$15,000 per seat give legal counsel case-law research access. Outside advertising-compliance counsel (Kelley Drye, Manatt) bills $400-$800/hr; multi-location operators retain at $50,000-$200,000+/yr. Substantiation-research firms (IMS Legal Strategies, expert-witness consultancies) charge $5,000-$20,000+ per substantiation memo. In-house VP-Compliance at $90,000-$200,000/yr bottlenecks once content scales past 50 pieces per week. Single-channel compliance tools (PerformLine, Hearsay, Smarsh, ProofPoint) enforce on one channel. General compliance suites (Compliance.ai, LogicGate, NAVEX) handle policy-level documentation, not per-claim substantiation with evidence linkage.
None of them maintain an operator-specific approved-claims library that AI content agents check programmatically at output time.
What success looks like
Every substantive claim your AI content agents produce gets checked against a per-operator approved-claims library at output time. Each approved claim is linked to substantiation evidence — clinical study citation, performance backtest, comparative-testing methodology, FDD Item 19 earnings methodology — plus an expiration date and per-vertical / per-jurisdiction conditions.
When an agent attempts a net-new claim, the borderline-routing skill sends it to legal for review. Approved claims get added to the library with substantiation linkage; rejected claims get added to the forbidden-phrase-library with rejection rationale.
When substantiation ages out — clinical study superseded, performance period ended, comparative methodology becomes stale — the claim auto-suspends across every content agent until re-substantiated.
Multi-brand portfolios get per-brand libraries. Multi-vertical operators get per-vertical extensions (FDA structure-function for supplements + cosmetics; FINRA suitability for financial; state-AG cannabis restrictions). Franchise operators get FDD Item 19 earnings-claim discipline. Westlaw stays accessible for ad-hoc case-law research; outside counsel still approves new claims — the skill makes their approvals operationally enforceable across every AI agent in the catalog.
How most operators solve this today
Six tiers of incumbent tools — none maintain a per-operator approved-claims library with substantiation evidence enforced at AI-content-output time.
Legal-research databases (Westlaw, LexisNexis)
$5,000-$15,000 per seat/year
Counsel-facing case-law research. Not operator-facing; no claim-library output; no AI-agent enforcement.
Substantiation-research firms (IMS Legal Strategies, Compass Lexecon, expert witnesses)
$5,000-$20,000+ per substantiation memo
Per-claim memo authored manually. Engaged for individual high-risk claims. Memos sit in legal team drives; AI agents have no access.
Advertising-compliance counsel (Kelley Drye, Manatt, Frankfurt Kurnit)
$400-$800/hr; retainer $50,000-$200,000+/yr
Approves claims and drafts substantiation memos. Approvals live in email threads. No structured library; no AI-agent enforcement.
In-house VP-Compliance / marketing-legal counsel
$90,000-$200,000/yr salary
Single reviewer bottleneck. Bottlenecks at 50+ content pieces per week. Approvals captured in Slack and Docs; no enforcement layer.
Single-channel compliance software (PerformLine, Hearsay Systems, Smarsh, ProofPoint)
$25-$65 per user/month or $30,000-$150,000/yr
Channel-specific (affiliate, financial-services social, email DLP). Does not cover full AI-content-production stack.
General compliance suites (Compliance.ai, LogicGate, NAVEX)
$20,000-$150,000/year
Policy-level documentation and workflow. No per-claim substantiation library; no content-output enforcement.
What changes when this is an agent skill
The Completions claims-allowlist skill maintains a per-operator approved-claims library — each claim linked to substantiation evidence (study citation, performance backtest, methodology), expiration date, and per-vertical / per-jurisdiction conditions. Every content-producing agent in the catalog checks each substantive claim against the library at output time.
Net-new claims trigger borderline-routing (loop 004) — legal reviews; approved claims get added with substantiation linkage; rejected claims feed the forbidden-phrase-library (loop 020).
Expiration alerts auto-suspend claims across every agent when underlying substantiation ages out. Per-vertical filtering composes with per-vertical-compliance-overlay (loop 001) — FDA structure-function claims for supplements + cosmetics, FINRA suitability for financial, state-AG restrictions for cannabis, FTC Franchise Rule Item 19 for franchise earnings claims.
Multi-brand and multi-jurisdiction operators get inheritance — corporate base library; brand / vertical / jurisdiction sub-libraries override or extend. Outside counsel still approves new claims; the skill makes their approvals operationally enforceable across every AI content agent. Westlaw, LexisNexis, and substantiation-research firms remain for ad-hoc deep research; the canonical library lives in the skill.
Agents that include this skill
Skills live inside agent rentals. To get this skill in production, hire any of the agents below — context-tuning at onboarding is included in the first month.
Brand-Spec Authoring + Maintenance Agent
Produces and maintains the canonical brand spec every content-producing agent's brand-voice gate enforces.
Early-adopter
$2,000–$3,500/mo
FAQ
- What is claims substantiation?
- The legal requirement that every advertising claim be supported by competent and reliable evidence before it is made. The FTC enforces against unsubstantiated claims; FDA enforces against unsubstantiated health claims; state AGs and class-action lawyers add additional risk. This skill maintains an operator-specific allowlist of pre-substantiated claims that AI content agents check against at output time.
- How is this different from a legal-research database (Westlaw, LexisNexis)?
- Legal databases give counsel access to case-law precedent for ad-hoc claim research. This skill is the operator's own claims library — built from the operator's substantiation evidence (studies, testing, performance data, FDD Item 19 methodology), enforced programmatically at AI-content-generation time.
- How is this different from advertising compliance counsel (outside law firm)?
- Outside counsel substantiates individual claims at $400-$800/hr. This skill productizes the substantiation library across every content-producing agent. Counsel still approves new claims; the skill makes their approvals operationally enforceable.
- What about FTC Franchise Rule earnings claims?
- Franchise operators making earnings claims must follow FDD Item 19 methodology. The skill supports franchisor-approved Item-19 earnings claims with substantiation methodology linkage; franchisee-level deviation triggers a borderline-routing review.
- What about FDA health claims for supplements and cosmetics?
- FDA distinguishes structure-function claims (allowed with substantiation plus disclaimer) from disease claims (require pre-market FDA approval). The skill's per-vertical filtering enforces the distinction at the language level.
- How are expired or invalidated claims handled?
- Every approved claim has a substantiation-evidence link plus expiration date. When evidence ages out — clinical study superseded, performance period ended, comparative testing methodology becomes outdated — the claim auto-suspends across every content agent. Re-substantiation re-activates it.
- Can different brands, verticals, or jurisdictions have separate libraries?
- Yes. Multi-brand portfolios get per-brand libraries; multi-vertical operators get per-vertical extensions; multi-jurisdiction operators get per-jurisdiction overrides. Inheritance hierarchy lets corporate set base claims; brand / vertical / jurisdiction sub-libraries override or extend.
- How does this compose with forbidden-phrase-library and per-vertical-compliance-overlay?
- Forbidden-phrase-library is the negative side — language the operator prohibits. Claims-allowlist is the positive side — claims the operator allows with substantiation. Per-vertical-compliance-overlay enforces regulator-required prohibitions. All three feed the brand-voice-gate which enforces at every content output.