About

A methodology platform — not a freeform AI agent.

Vigilarx makes research-security compliance defensible. Every claim that reaches a Person Report is anchored to a watchlist evidence row identified by name, version, and fetch-date. The institution is the named author of the certification — we provide the methodology and reference data, the institution adopts and signs.

What we are

A CelesiumAI Studio venture.

Vigilarx is operated by CelesiumAI, a venture studio that builds methodology-anchored AI products for regulated domains. We design the product, ship it under a single brand, and partner with practitioners in each domain to ensure the methodology is defensible.

What we are not

Not a private-label whitelabel.

Vigilarx is not a re-skin of someone else's compliance toolkit. It is a methodology platform built around the NSPM-33 four-pillar structure, anchored in named federal authorities (NSPM-33, NDAA §1286 / §1260H / §889 / §5949, UFLPA, EAR/ITAR, EO 14117, FCA), and developed against the operational requirements of practicing university research security officers.

Our position

Why this is hard, why we built it.

Research security and export control are two of the few federal compliance regimes where the institution is criminally and civilly exposed for getting it wrong, AND the underlying determinations are produced informally — emails, spreadsheets, PI memory, lunch conversations. The False Claims Act (31 USC §3729) and the federal false-statements statute (18 USC §1001) are unforgiving of informal records. When a finding is challenged, every line of the supporting dossier needs to map back to a named source whose version-of-record is preserved.

Freeform LLMs cannot do this. They generate text without anchoring it; their outputs are not auditable. Vigilarx's wedge is outcome-level: every claim that reaches a rendered dossier is anchored to a verifiable source row whose watchlist is identified by name + version + fetch-date, and the entire signed chain is preserved in the audit log. Tamper-evident SHA-256 attestation hashes confirm the artifact wasn't modified after signing.

This isn't a feature; it's the product. Everything else — the 6-section Person Report, the four-pillar Compliance Command Center, the seven-feature Export Control module, the cert PDF Exhibits A/B/C, the audit trail — sits on top of that defensibility foundation.

Mission

Make federal research-compliance defensible by default.

An RSO who signs an RSP certification, an Export Control Officer who approves a TCP, a PI who acknowledges the statutory awareness panel before signing — every one of them should be able to defend every claim in their signed artifact, line by line, against any federal review, without re-doing the underlying work, and without depending on an AI vendor who cannot produce the source rows behind their own model output.

Origin

Built around the asks of practicing RSOs.

Vigilarx was scoped from interviews with university research security officers facing the NSPM-33 implementation deadline. The 7-step workflow tour, the four-pillar tracker, the batch roster screening, and the FCA-defensible cert PDF all map to concrete operational asks raised in those conversations. We are deliberate about not over-claiming endorsement: customer institutions are named publicly only with explicit written consent.

Want to talk to us?

We'll respond within one business day. Pricing, security questionnaire, walkthrough on your roster, or methodology questions — same inbox.