Welcome.AIWelcome.AI
    The third verb: trust

    Trust, shown — not asserted.

    Welcome.AI helps organizations discover, evaluate, and trust AI companies. Trust is the hardest of the three, so we make it concrete: every claim carries its provenance, and the underlying data is public and machine-readable — for the buyers doing diligence and the AI agents acting on their behalf.

    Every claim shows where it comes from

    A trust signal is only as good as its source. Each disclosure on a company profile is tagged with one of three provenance levels, so a reader always knows what's a statement and what's been confirmed.

    Self-reported
    Stated by the company. The default for every claim until evidence is attached.
    Link-verified
    Backed by a public link a buyer can open and confirm — a policy, a report, a registry entry.
    Verified
    Confirmed by an independent auditor or certification body. A company can never mark its own claims verified.

    Whatever the level, one line never gets stripped: this data is self-reported by the company; not independently verified by Welcome.AI.

    What a trust profile covers

    Companies publish a structured profile across the dimensions buyers actually evaluate — the same surface a SafeBase or Vanta trust center covers, plus the AI-specific questions those tools skip.

    Data & training
    Whether customer data trains models, opt-out availability, retention, and residency.
    Foundation models
    The models in use, their providers, purpose, and hosting.
    Responsible AI
    Governance policy and a plain-language summary of how AI risk is managed.
    AI risk
    Disclosures mapped to frameworks like NIST AI RMF, ISO/IEC 42001, and the EU AI Act.
    Certifications
    SOC 2, ISO 42001, HITRUST AI and others — linked out so buyers confirm them at the source.
    Subprocessors
    Third parties in the data path and what each one handles.

    Built for AI agents too

    The same trust data is public and machine-readable. Agents doing research can fetch it directly — no scraping, no guessing — against an open schema.

    • trust.jsonThe full governance payload, per company.
    • agent.jsonThe machine-readable company profile, trust block included.
    • llms.txtA site-level guide for language models.
    • AI Entity SpecThe open schema these artifacts follow.

    Run an AI company?

    Publish a trust profile and let buyers — and their agents — verify you.