The neutral trust-data feed for AI tools.

You already discover the AI apps your customers use. Vannus is the enrichment layer underneath — independent trust scores, sovereignty & provenance, and a versioned change feed for the AI-tool category — delivered by API, so you don't have to stand up an AI-tool research team.

Built on the same catalog that powers Vannus: 287 AI tools, scored on a 9-dimension trust framework, with neutrality enforced in code — not popularity, not vendor self-attestation.

Why this is different from a review site or a vendor registry

Neutral, proven in code

A CI invariant forbids the scoring stack from importing any partner/affiliate module — break it and the build fails and the public attestation flips. We can show neutrality, not just assert it. G2 carries paid placement; this doesn't.

Independent, not self-reported

Vannus scores the tools against documented public criteria. This isn't a vendor self-attestation registry where the vendor fills in its own answers.

Provenance + freshness

Every record carries a source URL, a last-updated date, and a catalog version. A versioned /changes diff feed tells you exactly what moved since you last synced.

Who it's for

Shadow-AI & SaaS-management

You inventory the apps; we enrich each one with sovereignty, training-data usage, ZDR, compliance posture, and base model.

Vendor-risk / TPRM

AI tools are a new vendor category with no neutral source of record. Buy the feed instead of standing up research.

AI-governance / GRC

A neutral third-party reference for AI-vendor due diligence as the EU AI Act forces companies to inventory and vet their AI vendors.

The API

Read-only, versioned, JSON. Mounted under /api/catalog/v1. Per-customer keys.

Endpoint	Returns	Access
`GET /meta`	Catalog version, tool count, last-built timestamp, schema + neutrality-attestation URLs	public preview
`GET /schema`	The JSON Schema every record validates against	public
`GET /tools`	Paginated tool records (full scored record + provenance)	licensed
`GET /tools/{id}`	One tool — full scored record + provenance block	licensed
`GET /changes?since={version}`	Diff since a prior catalog version: added / removed / rescored, each with an evidence reference	licensed

The guarantees are the product

Every record and every release is held to invariants enforced by the test suite that gates the build. The CI gate is the integrity proof you can show your own risk team:

Neutrality — no paid-placement or affiliate signal in any record.
Schema-valid — every record validates against the published JSON Schema.
Stable IDs — a tool's ID never changes underneath you.
Monotonic versions — catalog versions only move forward.
No silent rescores — every score change carries an evidence reference in the change feed.
Provenance present — source URL + last-updated on every record.
Honest previews — the public preview is capped and flagged; no paywall bypass.

See the data

A real sample export — actual records from the live catalog, in the v1 schema (neutral, provenance-tagged, freshness-versioned). Not a price sheet.

{
  "meta": { "catalog_version": "2026.05.0", "tool_count": 287, ... },
  "sample_records": [
    {
      "id": "perplexity", "name": "Perplexity",
      "dimensions": { "...": "9 trust dimensions, 0–10" },
      "country_of_origin": "...", "is_us_controlled": true/false,
      "zdr_compliant": true/false, "training_data_usage": "...",
      "provenance": { "source_url": "https://…", "last_updated": "2026-05-20",
                      "catalog_version": "2026.05.0" }
    }
  ]
}

Download the full sample export (JSON) · Neutrality attestation · How neutrality is enforced

Become a design partner

We're working with a small number of design partners to shape the schema before general availability. Design partners get preview access and influence over the fields, in exchange for honest feedback. If you run a platform that discovers or governs AI tools, we'd like to talk.

Request design-partner access See how neutrality works

Honest scoping: the Catalog API is in a design-partner phase and is not yet generally available. v1 ships tool-level provenance (source URL + last-updated); per-field provenance is on the roadmap. Vannus is a neutral evaluation service, not a certification body — it makes no SOC 2 / ISO / BAA representations about itself, and trust scores are opinions formed from publicly disclosed criteria as of the stated date. See the Terms and neutrality methodology.