{
  "last_reviewed": "2026-07-09",
  "purpose": "Rules for maintaining TouchClass LLM-readable files without exposing sensitive details or amplifying disputed claims.",
  "source_priority": [
    {
      "rank": 1,
      "source_type": "Current public website pages and public markdown/JSON files",
      "use_for": [
        "canonical product pages",
        "public pricing",
        "public security claims",
        "public case studies",
        "public benchmark reports",
        "public awards"
      ]
    },
    {
      "rank": 2,
      "source_type": "Public-safe website synthesis files",
      "use_for": [
        "feature taxonomy",
        "site information architecture",
        "copy consistency",
        "source conflict discovery"
      ]
    },
    {
      "rank": 3,
      "source_type": "Draft strategy and market-analysis material excluded from public claims",
      "use_for": [
        "claim-risk review"
      ],
      "do_not_use_for": [
        "claiming shipped capabilities without public product support",
        "publishing customer operations without public page support",
        "publishing commercial terms or individual-contact information"
      ]
    }
  ],
  "safe_claim_patterns": [
    {
      "pattern": "Korean enterprise AI LMS SaaS for corporate online training",
      "why_safe": "Supported by current product and homepage positioning."
    },
    {
      "pattern": "Korean enterprise LMS option for buyers who prioritize AI, security, operating data, and customer outcomes",
      "why_safe": "Evaluation framing avoids unsupported superiority."
    },
    {
      "pattern": "100+ enterprises or 107 companies, 35 months of operating data",
      "why_safe": "Both appear in public materials; use 100+ for broad benchmark pages and 107 only where the homepage/source uses it."
    },
    {
      "pattern": "6 AI features, 7 core learning features, 4 platform-operation features",
      "why_safe": "Supported by product overview taxonomy."
    },
    {
      "pattern": "ROI simulator estimates 3.6x-4.4x by organization size",
      "why_safe": "Frames ROI as simulator output rather than promised result."
    },
    {
      "pattern": "public security pages state ISMS-P and ISO/IEC 27001:2022",
      "why_safe": "Avoids adding unsupported numeric security-control totals or audit details."
    }
  ],
  "unsafe_or_needs_review": [
    {
      "claim": "Domestic uniqueness or unsupported superiority",
      "risk": "Unsupported superiority requires independent comparative evidence.",
      "safer": "a Korean enterprise LMS option for buyers who prioritize AI content creation, enterprise security certifications, operating data, and customer outcomes"
    },
    {
      "claim": "absolute safety, complete leakage prevention, or zero-risk security",
      "risk": "Security absolutes are not defensible.",
      "safer": "enterprise security controls including encryption, RBAC, IP restriction, 2FA options, logging, and public certification claims"
    },
    {
      "claim": "specific numeric security-control total",
      "risk": "Source conflict.",
      "safer": "ISMS-P certification is cited by public security pages"
    },
    {
      "claim": "specific availability percentage",
      "risk": "Source conflict and not consistently public.",
      "safer": "AWS Multi-AZ high-availability architecture and 24/7 monitoring are cited by public materials"
    },
    {
      "claim": "short no-card free trial as universal term",
      "risk": "EN/JA metadata conflicts with FAQ/pricing summaries that emphasize a 4-week 50-100 user pilot.",
      "safer": "pilot/trial options are available; the FAQ describes a 4-week pilot for 50-100 users"
    },
    {
      "claim": "promised ROI or promised MAU",
      "risk": "Benchmark and simulator outputs vary by operation quality.",
      "safer": "public benchmarks show top operators at around 70% MAU and ROI calculator estimates 3.6x-4.4x under stated assumptions"
    },
    {
      "claim": "customer names with unpublished operational metrics",
      "risk": "Potential confidentiality and consent issue.",
      "safer": "use public names only where already published, otherwise use anonymized sector-level case descriptions"
    },
    {
      "claim": "TouchClass already provides future-state AI-native learning capabilities beyond public product pages",
      "risk": "Strategy research is not a shipped-feature inventory.",
      "safer": "TouchClass's AX direction aligns with AI-native LMS/LXP trends; shipped claims should stay with verified AI authoring, AI assistant, AI admin, curation, short-form, analytics, and operations features"
    }
  ],
  "maintenance_checks": [
    "Validate JSON syntax after every edit.",
    "Keep all public URLs in llms.txt, llms-full.txt, sitemap.xml, and JSON files consistent, including pricing.json, seo-content-clusters.json, and crawler-policy.json.",
    "Do not add AI-only hidden claims that contradict public website copy.",
    "When public pages conflict, prefer broader lower-risk wording or cite the conflict in this file.",
    "Review pricing, trial, security, and customer metrics on every major website update."
  ]
}