Inventory
Nobody knows how many agents there are
Custom GPTs, projects, connectors and code agents appear without a record. The last survey is already wrong.
Observe · Compliance teams
What AI is in use? What can it reach? Who owns each agent? Lustena keeps the inventory, the limits and the decision history current, and exports an evidence package mapped to GDPR, the EU AI Act, ISO/IEC 42001, NIST AI RMF, AIUC-1 and LGPD.
The AI policy is a document, the inventory is a spreadsheet, the approvals are in email. None of them is current when the questionnaire arrives.
Inventory
Custom GPTs, projects, connectors and code agents appear without a record. The last survey is already wrong.
Decisions
Who allowed the sales GPT to read the CRM, when, and under which limit? The answer is an email thread, if it exists.
Lawful basis
Reading employee prompts to protect data creates a privacy problem of its own. Evidence has to be content-free by default.
Each agent carries an owner, a team, a purpose, its connectors, its limit and its findings. Declared and observed stay distinct, so the gap is visible instead of hidden.
Agents
Every agent this workspace knows about, declared by a person or observed in use, with its owner and connectors.
Inventory · 5 rows
| Agent | Owner · connectors | Status |
|---|---|---|
| GPT · Proposalsana@acme.com | ana@acme.comGmail, Drive | Declared |
| Claude · Finance projectjoao@acme.com | joao@acme.comSheets | Observed |
| Support bot (LangGraph)platform team | platform teamZendesk API | Declared |
| Copilot Studio · HR FAQunknown | unknownSharePoint | Observed |
| Claude Code · repo-assistengineering | engineeringMCP: GitHub | Declared |
Selected · GPT · Proposals
Owner
ana@acme.com · Marketing
Limits
v3 · tools, hosts, data
Last seen
2m ago · chatgpt.com
Every limit is versioned with who changed what and when. Alerts record the dimension crossed, the observed value and the allowed one. This is the decision record ISO 42001 asks for.
Limit history
Every change is a new version, attributed and dated. Previous versions stay readable.
Versions · 3
Alerts · limit v3 · 24h
data class: personal · observed 3 × card number · allowed: none
paste · chatgpt.com · Oct 14 · 02:41 · webhook → #sec-alerts
Findings are content-free by default. Redacted capture is opt-in, requires a written lawful basis and a retention period, is encrypted at rest and audited on every read. The evidence package maps all of it to the frameworks you name.
Evidence package
One document for your auditor, mapped to GDPR, EU AI Act, ISO/IEC 42001, AIUC-1.
Contents · 5 sections
| Section | GDPR | AI Act | 42001 | AIUC-1 |
|---|---|---|---|---|
| Inventory with owners14 agents · 5 teams | ||||
| Limits and versions9 limits · 23 versions | ||||
| Findings with lifecycle29 open · 112 closed | ||||
| Scan reports4 endpoints · OWASP LLM Top 10 | ||||
| Decisions37 changes, attributed |
Redacted capture · Marketing
lawful basis: recorded · retention 180d · reads audited
findings: content-free by default · values replaced by category
Questions
GDPR, the EU AI Act, ISO/IEC 42001, NIST AI RMF, AIUC-1 and LGPD today. Named standards are references; Lustena prepares the evidence, independent assessors certify, and Lustena does not issue certifications.
Not by default. Opt-in redacted capture stores text with every detected value replaced by its category, with a lawful basis recorded and every read audited.
Yes. The evidence package includes the inventory, limits, findings and decisions, and is shared with your auditor as one document.