Know what AI your organization uses

Discovery findings, the spreadsheets you already keep and requests from colleagues all bring AI systems into one AI Registry. Each one starts as a draft, gets an owner and goes through a review before it counts as approved.

Request a demo See what happens after approval

In one week 43 AI systems reach the AI Registry: 39 found by discovery, three imported from a spreadsheet and one submitted by a colleague. Seven of them are shown. Each gets an owner and goes to review; the Registration Manager approves five, one waits for review and one stays in draft because its intended use is missing.

AI Registry1,048 approved1,049 approved1,050 approved1,051 approved1,052 approved1,053 approved

All 43 this weekFound by discovery 39Imported from a spreadsheet 3Submitted by a colleague 1

AI systemArrivedOwnerRegistration

Customer reply agent 4Marketing · found by discoveryTue 15 SepMarketing team leadDraftPending reviewApproved
Meeting notes agent 7Claims · found by discoveryTue 15 SepClaims analystDraftPending reviewApproved
Meeting notes agent 5Customer service · found by discoveryTue 15 SepCustomer service analystWaiting for reviewDraftPending review
Underwriting copilotUnderwriting · imported from a spreadsheetWed 16 SepUnderwriting leadDraftPending reviewApproved
Commercial pricing modelUnderwriting · imported from a spreadsheetWed 16 SepPricing actuaryDraftPending reviewApproved
Lead scoring modelSales · imported from a spreadsheetWed 16 SepSales operationsIntended use missingDraft
Supplier risk scorerProcurement · submitted by a colleagueThu 17 SepProcurement analystDraftPending reviewApproved
and 36 more found by discovery this week

Your AI inventory lives in too many places

AI arrives through procurement, through low-code licences and through code that a team shipped last month. A spreadsheet register falls behind within weeks, and nobody is sure who owns half of the rows.

The first thing an auditor asks for is the inventory, and the second is the name of the person who answers for each system. Both answers should come from one place, and they should be current.

Find the AI nobody registered

Your security and platform tools already see AI that never reached the register: agents built on low-code platforms, models running in cloud accounts, assistants a team set up on its own. But a finding in a security console has no owner and no review. Modulos compares what these sources report with the AI Registry and opens a draft for anything missing, with where it was found and when. Access to every source is read-only, and nothing is approved on its own.

Read about shadow AI governance

AI Registry / Drafts

Forecast helper agent 2Draft

Source: Found by discovery · 4 Sep 2026
Found in: The Claims workspace of a low-code agent platform
Built by: A claims team lead
Owner: Not assigned
Intended use: Not given

The owner completes the record and submits it for review. Nothing is approved on its own.

Found by discovery in the last 7 days · 39

Policy Q&A agent 2Claims · 16 SepDraft
Vendor check agent 7Claims · 16 SepDraft
Report summary agent 5Customer service · 16 SepDraft

and 36 more

Register an AI system with Scout beside you

Anyone can register a new AI system or one the team already uses. The wizard asks for the system and its details, then the EU AI Act context and the sources, then the frameworks that apply. Scout sees what you have entered so far, so it can help with any field.

Economic value can start as a rough estimate, and the owner refines it later.

AI Registry / New AI system

1AI System & Details
2EU AI Act & Sources
3Frameworks

Name: Supplier risk scorer
Owners: Procurement analyst
Intended use: Scores new suppliers for financial and ESG risk before onboarding
Current state: Pilot
Lifecycle stage: Design and development
Run scope: Internal
Economic value per year: €400,000Rough estimate

Next: EU AI Act & Sources

Scout

You

Which run scope fits? Only our procurement team uses it.

Scout

Internal fits, because only the procurement team uses the scorer. If suppliers will see their scores, choose Customer-facing. For economic value a rough order of magnitude is enough for now, and you can refine it later.

Send each new AI system through a review

A new AI system starts as a draft. Before it can be submitted, the name, the intended use, the current state and the run scope must be filled in. A Registration Manager then approves it or rejects it with a reason, and the reason goes back to the person who submitted it.

Current state and registration status are separate fields, so a reviewer can see a system that is already in use but still waiting for a decision.

Supplier risk scorerPending review

Procurement · submitted by a procurement analyst

Name: Supplier risk scorer
Intended use: Scores new suppliers for financial and ESG risk before onboarding
Current state: Pilot
Run scope: Internal

The four fields above are required before a draft can be submitted.

Start from the spreadsheet you already keep

Import your current inventory from a CSV file. Every row is checked before anything is saved, so missing fields and unknown owners show up in the preview and not in the register. After the import, each role does its own part of the work.

underwriting-inventory.csv · preview6 rows checked

Row	AI system	Unit	Check
12	Claims triage assistant	Claims	Ready
13	Damage photo classifier	Claims	Ready
14	Lead scoring model	Sales	Intended use missing
15	Invoice matching model	Finance	Ready
16	CV screening assistant	HR	Owner not found
17	Code review agent	IT	Ready

Nothing is saved until you import.Import 4 AI systems

Owner: Answers for the system and keeps its record current
Editor: Edits the record for the owner
Registration Manager: Approves or rejects each submission, with a reason
Viewer: Reads the record and its history

Put a price on what discovery finds

In the sample portfolio, the largest single source of risk is an agent that discovery found on 4 September. A claims team lead built it but never registered it. In the register it joins the weekly Fermi run, and an AI system outside the register never reaches that run.

The same record carries the lifecycle stage and the frameworks chosen in the wizard, so governance and compliance start from it.

See how the portfolio is priced Follow a system through its lifecycle

Found by discovery · Claims

Forecast helper agent 2

Built by a claims team lead · found 4 September

€3.0M a year across 30 threats

Sensitive information disclosure€976K
Degraded accuracy€251K
Excessive agency€169K

Weekly Fermi run, Monday 21 Sep · sample portfolio

Questions about discovery and registration

Can we start with a spreadsheet?

Yes. Import your current inventory from a CSV file. Every row is checked before anything is saved, and each imported system then goes through the same registration and review as any other.

How does Modulos work with our discovery tools?

Your security and platform tools see the environments they cover. Modulos compares what they report with the AI Registry and opens a draft for anything missing, so a finding becomes a record with an owner, a review and a lifecycle stage. Access to every source is read-only.

Can employees register AI that is already in use?

Yes. The wizard records the current state of a system (proposed, pilot or already in use) apart from its registration status, so reviewers can see a system that is already in use but still waiting for a decision.

Does registering a system mean it is approved?

Every new AI system starts as a draft. Once the required fields are complete it goes to review, and a Registration Manager approves it or rejects it with a reason that goes back to the person who submitted it.

Bring the records you have and the gaps you know about

We start with where AI runs in your organization today and how your team hears about new systems. Then we walk through registration, review and what approval sets in motion.

Request a demo Read about shadow AI