CEDX Desk · Customer Service
Desk is ticketing built around a forecast: every open ticket carries a breach score and the signal that drove it, the queue lead sees the 51 nobody has claimed, and the deflection panel prints its own economics.
The deflection panel does its arithmetic in public — $0.99 against $6.40 per resolved outcome, 85% lower, printed next to the SLA ring it relieves. Nothing on this page asks you to take a number on faith.
Runs on demo data. The workspace is the software's own demo tenant, not a customer.
What it is
The queue sorts oldest waits first: T-1446, a password-reset ticket, sits at 19.8 hours with its first-response goal 17.8 hours past. WAIT and FRT LEFT are columns, not reports — every row carries both, next to priority, status, assignee and org.
The triage engine scores all 201 open tickets; 112 sit at breach ≥55% and 140 have already breached — three separate counts, none of them hidden. Each row names its top signal: “queue wait 3.2h vs FRT goal 1h” is a reason you can act on.
The knowledge screen counts what agents actually reuse: 1,757 uses in 30 days, 435 deflections as self-serve hits, and 24 items flagged stale — including the reply guidelines macro agents ran 66 times last month.
Product tour
Not a mockup and not a concept deck. This is what opens at /app/desk.
01 — Overview
The created-vs-solved chart carries the week's events as markers — an IdP cert rotation window in W1, a VIP policy going 24×7 in W12 — so volume moves have named causes. The SLA ring says 30% of the open queue is inside its clock, and the breach list below it is sorted by risk, 95 at the top.
02 — Tickets
All 320 tickets in the workspace with the tabs that matter: open 201, unassigned 51, breached 140. The top of the list is a password-reset MFA ticket at 19.8 hours — the sort order is the argument for working it first.
03 — Triage
Every open ticket scored, the riskiest first: T-1042, an SSO login loop after an IdP rotation, sits at 95 with its FRT goal already past. The scoring model and the SLA simulator are tabs on the same screen, not a separate product.
04 — Knowledge
Eighty items — 39 articles, 41 macros — with uses-30d, status, author and updated date on every row. The most-used macro, the chat greeting, ran 88 times in 30 days; the stale filter isolates the 24 that are quietly going wrong.
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 unassigned column and the shift handoff. The LED board version of this job is the photograph; the Desk version is the breach forecast read before standup.
unassigned · 51
Keeps the 80-item library worth reusing: retires the 24 stale items, writes against what the queue actually asks.
stale · 24Watches the forecast queue and the duplicate clusters, and retunes the scoring model when the signals drift.
forecast ≥55% · 112The 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
Every channel lands in one queue — 320 tickets in the demo workspace, 201 of them open, duplicates clustered into 27 groups.
The triage engine scores each open ticket for breach risk and prints its top signal — 112 sit at ≥55% with the reason attached.
Fifty-one tickets sit unassigned. The queue lead works the forecast order, not the arrival order.
Knowledge does the quiet work — 435 self-serve hits in 30 days, with the 24 stale items flagged before they start lying.
One record
Desk is not a mailbox with a database behind it. The same demo estate shows up on every screen — same workspace, same Ask AI button, same record.
The articles Desk's knowledge screen reuses 1,757 times a month live their public life in the help centre — with the same stale flags.
CEDX Live ChatSame Northline demo floor: 42 visitors waiting in chat sit next to Desk's 201 open tickets, and long threads become tickets.
CEDX VoiceVoice's dropped and abandoned calls — 42 in its window — are the callbacks that arrive here as tickets.
CEDX CopilotAsk AI — ⌘J — sits on every Desk screen. The Copilot console is where those prompts are versioned, scored and traced.
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 queues, your SLA clocks
Desk with the rest of service
Questions
Yes. Every screenshot is a capture of the running build and you can open the same build at /app/desk. It runs on demo data — the workspace name on the chrome is the demo tenant, not a customer.
Every open ticket gets a breach score and a top signal — queue wait against the first-response goal, age against response, and similar. The scoring model is a tab on the triage screen itself, next to the SLA simulator, so you can read what drives the number.
They stay counted. The tickets screen has a breached tab — 140 in the demo — and the overview ring prints breached next to within-SLA rather than rolling them into a single green number.
From the demo tenant's own panel: 61% AI resolution rate with estimated costs of $0.99 and $6.40 per outcome printed beside it. Those are the demo workspace's economics, shown because the screen shows them — not a benchmark for yours.
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 check what your own unassigned column would look like.