CEDX Expenses · Finance
965 lines booked this month, $38K waiting on approval, 150 policy flags and 64 missing receipts — the overview reads like the month-end conversation you were going to have anyway, except it is already priced.
A flag is not a vibe here. Each blocked line prints its reason — “professional services hard-blocked without PO” — next to the amount it stopped.
Runs on demo data. The risk board's own math — amount × risk% = expected $ — is printed on the screen.
What it is
208 lines scored with the formula on screen: amount × risk% = expected dollars. Lumen Telecom at $2.5K and 86% risk expects $2.1K of violation; the policy column says Block or Flag, and the ≥55% / ≥70% chips cut the board down to what needs a person.
47 signals, each named by shape: suspect document, merchant spike, late or high-fare travel. Each row carries a deviation percentage — 84% off peer for a suspect Beacon Legal document, 71% for a Printfly merchant spike — so the queue sorts by how strange, not how recent.
120 receipts in the demand queue with a 37% match rate. Nine suspect documents worth $5.3K are held for forensics, and the authenticity column says why: “Reviewed · edit metadata” on one, “Suspect · edit metadata +1” on another.
Product tour
Not a mockup and not a concept deck. This is what opens at /app/expenses.
01 — Overview
MTD spend $21K across 965 lines, $38K pending approval over 51 reports, 150 policy flags of which 42 are high-risk. Alerts arrive priced: Beacon Legal blocked at $3.2K for a hard-block without PO, and 49 approvals past the 5-day SLA holding $28K.
02 — Policy risk
Risk board 208, sorted by expected dollars. The top rows are blocks, not warnings: Lumen Telecom travel at 86%, Printfly at 85%, Beacon Legal professional services at 79%. Two simulation levers show $24K of expected loss and $25K saved against baseline.
03 — Anomalies
Suspect documents, merchant spikes and late high-fare travel, each with owner, department, deviation and amount. The heaviest cluster is suspect documents in Sales at 70–84% deviation — the kind of thing a spot-check finds a quarter late.
04 — Receipts
The receipt queue shows why the match rate is 37%: City Parking at $42 missing twice, a $987 Willow Cowork receipt matched but reviewed for edited metadata, partial matches on Skyline Airlines and Fleet Card Co. Age runs to 17 days on the oldest.
Who runs it
Roles, not references. We have no named customers yet, so nobody in these photographs is quoted, credited or claimed as one.
Photographs the receipt at the table and is done. The matching, chasing and awkward reminder emails happen in the queue, not at dinner.
37% match · 32 missing
Works the 49 approvals past SLA from a board that already knows the risk score, the policy and the expected dollars — instead of a thread of forwarded PDFs.
51 reports · $38K pendingOwns the anomaly board and the suspect-document hold pile, and answers for the policy health score at month-end.
47 anomalies · 9 docs heldThe 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
Spend lands by card or reimbursement — the overview plots both paces side by side so a reimbursement bulge is visible before it becomes a backlog.
Every line is scored against policy: amount × risk% = expected dollars. Blocks stop, flags queue, and the reason rides on the row.
Missing receipts become a demand queue with age in days; 32 are outstanding worth $19K, and suspect documents are held rather than paid.
Approvals clear from a board that shows risk and policy together — 49 are past the 5-day SLA right now, priced at $28K so the cost of waiting is visible.
One record
One demo workspace means the merchant names and amounts on these expense lines are the same ones the rest of the finance apps see — no second truth to reconcile against.
The card-and-request side of the same book: budgets, vendor concentration and the agent that auto-clears routine spend.
CEDX PayablesWhere a hard-blocked-without-PO line would have become a bill — the early-pay, duplicate and approval screens run on the same vendors.
CEDX BooksThe ledger the approved lines post to, and the finished reference page for how this estate presents its evidence.
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 policy, your merchants
Expenses 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/expenses. It runs on demo data.
What the screen says: amount × risk% = expected dollars. The inputs sit on a scoring-model tab next to the board, and the two simulation levers show the expected loss moving from $50K to $24K against a $25K baseline saving.
Four named shapes, visible on the board: suspect document, merchant spike, late or high-fare travel, and peer-group deviation. Each row prints its deviation percentage — the queue sorts by how strange a line is, not when it arrived.
Because that is the state of the demo book, and the screen shows it instead of hiding it: 120 receipts in the queue, 32 missing worth $19K, 9 suspect documents held. Your number would be your own; the demo's job is to show the mechanics.
It is held, not paid. The receipts screen shows nine documents worth $5.3K in the forensics hold, with the authenticity column recording why — edited metadata is called out explicitly.
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 count your own missing receipts and compare the view.