DNLA Registry

In development

Most organizations have no real inventory of how many AI systems they're running, who owns them, what data they touch, or what happens if one of them fails. DNLA Registry is a central, living record of exactly that.

The problem it solves

Shadow AI is the default, not the exception

AI systems get adopted faster than they get inventoried. A team spins up a chatbot, someone wires a model into a spreadsheet macro, a vendor quietly ships an "AI-powered" feature inside a tool you already pay for, and six months later, nobody in security, legal, or leadership can produce a complete list of what's actually running, on what data, with what permissions. That gap is where incidents, compliance failures, and unmanaged risk accumulate silently.

A team quietly wires an LLM into a customer-facing workflow with no security review, no owner, and no one else in the company aware it exists

A vendor changes the underlying model behind an integration your company depends on, and nobody notices until behavior shifts

An auditor or regulator asks for a full inventory of AI systems in use, and the honest answer is nobody actually knows

A system that has been quietly failing for months resurfaces during a Health Check, because no one owned reviewing it

What it tracks

One record per system, kept current

Every AI system in the organization
Business and technical owner
Vendor and model
Purpose of use
User groups
Data sources
Sensitive information handled
Risk level
Last review date
Renewal date
Model version
Vendor dependency
Approvals and exceptions
Incidents and faults

Registry is designed to sit alongside QAi and Governance Readiness; every audit and every governance engagement feeds it, so the organization ends up with one place that knows what AI it's actually running, instead of a fresh discovery exercise every time someone asks.

Not sure how many AI systems you actually have live?

Get in touch