Vaughn Harvey & Co

the AI governance platform · Registry and marketplace

One registry for every AI asset, so approval keeps pace with adoption.

Why now

Two waves hit the same governance team. Every licensed platform added AI assistants, copilots and agents in the last year, each reviewed from zero. Employees now build agents, skills and plugins in Copilot, ChatGPT and Claude, with no path to review at all. Same reviewers, same checklists, same monthly committee: three to six months per approval, and most AI use unseen.

How it works

  1. 1

    Register.

    Every AI asset, from a prompt to an agent to a vendor platform, with an owner, a tier, a status and its dependencies, discovered from the platforms' own consoles.

  2. 2

    Declare and tier.

    The owner declares data, autonomy, impact and review; the firm's own versioned rules compute the tier, and every tier names its rule.

  3. 3

    Match a pattern.

    Published patterns for MCP services, skills and plugins, third-party solutions and access turn repeat reviews into an exception check.

  4. 4

    Review in parallel.

    Cyber, information security, data, risk, legal and architecture work from one record with a clock on each; the committee decides the high tier only.

  5. 5

    Publish to the marketplace.

    Approved assets go live in the firm's marketplace: install in one click, with their conditions, identity and kill switch attached.

  6. 6

    Evidence.

    An exam pack for every decision; conditions tracked to closure; re-review on a version, permission or publisher change.

The registry is the inventory, the patterns are the speed, the marketplace is the reuse, the exam pack is the proof.

The marketplace

One governed catalogue of approved skills, agents, connectors and vendor AI. A skill built in one line is registered once, reviewed once, and reused across the firm, with every line seeing what is approved and what it is approved for.

Built on your standards, not ours

The platform ingests the firm's own control frameworks (governance standard, cyber control standard, AI governance standard) and links each control to NIST, NYDFS and the regulations that apply to you. Tiering is rules-based: certain criteria, such as agent identity management, must be satisfied or the asset does not pass. The platform is the registry; approval workflows can run through the GRC tooling you already have.

Design-partner engagements

Register one line's assets, produce the first exam pack, and shape the product with us. Available now for a small number of firms.

Register one line's assets and produce the first exam pack.

Ask about a design-partner engagement