CEDX Access · IT & Security
CEDX Access scores every grant for risk, ages every standing privilege, and puts a decision on every review row: keep, revoke, or convert to just-in-time — with an AI recommendation you can overturn.
The riskiest grant in the demo is a break-glass account at 94, used this morning with no ticket claim attached. The root-caused alert names it. That is the point of a risk score.
Runs on demo data — Northline is the software's sample tenant, not a customer.
What it is
The grant table carries principal, resource, role, level, type and risk: the break-glass account on the ledger at 94, a service account approving finance at 92, a contractor's deploy grant at 91. Unused for 90 days is a counted state — 64 grants.
The review board is 95 items with a decision per row: revoke the break-glass, keep the app-admin, convert the vault grant to just-in-time. The AI recommendation sits next to the decision — deny 97% — so agreement and disagreement are both recorded.
43 open requests, 9 of them high risk. Each shows requester, resource, role, duration and the AI recommendation — deny the standing payroll admin at 95, convert the deploy grant to 24 hours, approve the scoped one at 74% confidence.
Product tour
Not a mockup and not a concept deck. This is what opens at /app/access.
01 — Overview
281 standing grants, 43 requests awaiting decision, 9 high-risk, an access-health ring at 49. The root-caused alerts are the week's arguments: a contractor keeping deploy-operator unused since May, an SoD conflict where one person holds approver and deployer.
02 — Reviews
10 campaigns at 41% completion, 29 items pending attest. The board sorts by risk, and every decision records what the agent thought — the 37% overturn rate is a feature: humans disagree measurably.
03 — Grants
Filter to standing, sort by risk, and the ledger reads itself: break-glass at 94, service accounts at 92 and 80, contractor roles at 91. 13 grants belong to non-human agent principals, and they are labelled.
04 — Requests
A request names its duration up front — standing, 24 hours, 30 days — and the AI recommendation prices the ask. Convert-to-JIT at 83% confidence is the most common middle path in the demo queue.
Who runs it
Roles, not references. We have no named customers yet, so nobody in these photographs is quoted, credited or claimed as one.
Owns the break-glass accounts and the JIT sessions — a 94-risk grant used without a ticket is their first conversation of the day.
JIT sessions · 45
Runs the quarterly recerts: 95 items, decisions per row, and the 29 pending attests chased before the campaign closes.
review completion · 41%
Decides the 43 open requests with the risk score and the AI recommendation on the same row — and their overturns are counted, not judged.
overturn rate · 37%The shape of it
Every figure below is legible in the captures above. Nothing here is a projection of your estate — it is the state of the demo data.
How it runs
Access starts as an ask with a duration — standing, 24 hours, 30 days — and a risk score before anyone decides.
The approver sees the risk, the AI recommendation and their own history — overturns run at 37%, and the product treats that as signal.
The grant works, and the clock starts: last used is on the row, and 90 idle days moves it to the review candidates.
Campaigns turn the ledger into decisions: keep, revoke, convert to just-in-time. Completion is 41% and the board says so.
One record
What you can reach is a join of who you are, what you hold, and what it can touch — the secret, the event, the control. One record, four readers.
The 22 grants against prod/payments are this ledger's rows — Vault counts what they can read, Access counts whether they should exist.
CEDX ComplianceLogical-access controls test against this ledger: review completion, stale standing grants and the break-glass log are dated evidence.
CEDX SIEMBreak-glass use and role grants land in the event stream — the 08:53 ledger access with no ticket is an alert there and a finding here.
CEDX EndpointPrivilege follows the person to the device: a 94-risk grant on a 95-exposure laptop is the compound number both products show.
Limits
Finding this out on the third call is worse for you than reading it here, and worse for us.
Start
Open it right now
Your grants, your reviews
Access with the rest of it
Questions
Yes. Every screenshot is a capture of the running build and you can open the same build at /app/access. It runs on demo data — Northline is the sample tenant.
The scoring model is its own tab, and every row names its top signal — unused for 90 days, critical resource, standing where JIT would do. The break-glass account at 94 is the demo's honest example.
The middle decision between keep and revoke: the grant stops being standing and becomes time-bound — 24 hours, 7 days — issued on request. 58 JIT grants exist in the demo estate, with 45 sessions active right now.
They are recorded next to every decision, and the overturn rate is tracked: 37% over 30 days in the demo. A recommendation you can measurably disagree with is more useful than one that hides its mistakes.
No certification has been issued. What we can evidence about hosting, encryption, tenant isolation and production access is written up on the security page.
Live build, demo data, no card. Then ask which of your standing grants would survive a 90-day unused filter.