Use case · Finance
The close, without the week.
Cedric pulls the numbers, matches what should match, and flags what does not. You review the exceptions instead of assembling the spreadsheet.
$50 in free credits when you sign up. No card needed.
- Most of the monthly close is matching records between systems that were never designed to agree.
- Cedric reconciles line by line across your billing, banking, and accounting tools, and surfaces only the mismatches.
- Every figure it reports comes with the underlying records attached, so the board pack can be checked line by line.
- It never moves money. Payments, refunds, and journal entries are proposed and wait for your approval.
- The recurring parts (the weekly cash view, the monthly variance note) run on a schedule and post themselves.
The close is a matching problem
The processor, the bank, and the ledger each hold a slightly different version of the same month, and comparing rows until they agree eats the first week of every month.
Only the mismatches need a brain: the payout short by a conversion fee, the refund in the wrong period. Everything up to finding them is mechanical, and that is the part Cedric does.
Reconciliation
Line by line, before you have opened the spreadsheet.
Grace Adeyemi9:14 AM
@Cedric reconcile last month's Stripe payouts against the bank and flag anything that does not match.
CedricAPP9:15 AM
212 payouts, 209 matched to the penny. Three exceptions: a $412.60 payout short by $8.15 (currency conversion on 14 March), one duplicate refund of $99 issued twice to the same customer, and a $2,340 payout that has not landed in the bank yet (initiated 31 March, still in transit). Working file attached with the matched lines and the three exceptions on their own tab.
Every line checked
Not a sample and not a summary. Each record is matched against its counterpart, and the ones that agree are set aside.
Exceptions with reasons
A mismatch arrives with a probable cause and the underlying records, not just a red cell for you to investigate from scratch.
The working shown
The output is a file you can open, check, and file. Every figure traces back to the records it came from.
Where the numbers come from
Every figure arrives with its sources: when Cedric says revenue was up 12%, the file behind it holds the transactions that produced the number. A number with its records attached can be checked in a minute and corrected once, at the source.
Access is scoped and revocable, and reading your billing data never implies touching it: anything with consequences waits for a person, whatever the access technically permits.
Approval
It never moves money on its own.
CedricAPP10:02 AM
The duplicate $99 refund needs reversing. I have prepared it: reverse the second refund issued 14:22, 12 March, customer reference 5817. Approve?
Oliver Brandt10:03 AM
Hold that one, I want to check with the customer first.
CedricAPP10:04 AM
Held. Nothing sent. It stays on the exceptions tab as open, and I will leave it there until you say otherwise.
Getting to a real close
In the order that works. Each step is useful on its own, so you can stop at any point and still be better off.
- 01
Start with one reconciliation, not the whole close
Pick the pair of systems that wastes the most time (usually the payment processor against the bank) and get that loop right before adding anything else.
Resist doing the whole close in week one. One reconciliation that people trust is worth more than six they double-check by hand.
- 02
Connect the systems of record
Your billing platform, your bank feed, your ledger. Scoped access per tool, revocable from the dashboard whenever you want.
- 03
Run it against a month you have already closed
The best first test is a month whose answer you know. Compare what Cedric finds to what you found. Disagreements are informative in both directions.
- 04
Move the recurring reports onto a schedule
The weekly cash position, the monthly variance note, the ageing debtors list. These are the same query every time, which is exactly what should not need a person.
- 05
Let it prepare the board pack
Once the underlying numbers are trusted, assembling the pack is formatting. Cedric drafts it with the records attached; you write the narrative that only you can write.
How teams close today
All four of these are real approaches. Three of them are fine at some sizes.
| Spreadsheets | A custom script | Outsourced bookkeeping | Cedric | |
|---|---|---|---|---|
| Setup effort | None, it already exists | Weeks of engineering | Onboarding and handover | Connect the tools, one afternoon |
| Handles a new edge case | Someone edits a formula | A code change and a deploy | An email thread | Ask in the channel, in words |
| Breaks when a schema changes | Silently | Loudly | Not your problem | It says what changed |
| Traceable to source records | If the person was careful | Yes | On request | Yes, attached by default |
| Who can run it | The person who built it | Whoever can run the script | The provider | Anyone in the channel |
| Moves money without asking | No | Possibly, if written to | Sometimes | Never, approval required |
Where this breaks
Cedric is not your accountant, and it is not an audit. It prepares, reconciles, and flags. Sign-off, judgment on treatment, and anything with a regulator attached stay with qualified humans.
Garbage in stays garbage. If the same customer exists three times in your billing system under three spellings, reconciliation will surface the mess. Surfacing it is useful. It is not the fix.
Period cutoffs need stating. Ambiguity about which month a late-arriving transaction belongs to is a policy question, not a data question. Tell it your policy once and it will apply it consistently.
It will not silently guess. When something does not reconcile and the cause is not determinable from the records, it says so rather than forcing a match. Expect open items, especially in the first month.
FAQ
Only with an explicit approval, every time. It prepares the action, states exactly what will happen, and waits. Anything that sends, changes, or spends is gated the same way, and holding an action is always an option.
Cedric connects to the tools your team already uses, with permissions you grant per tool. If your finance stack has an interface Cedric can reach, it can read the records there and write back only what you approve.
Every figure Cedric reports comes with the records that produced it. Checking one means opening the working file, not rebuilding the calculation, and a disagreement points at a specific line.
No. It removes the matching, the collating, and the report assembly, which is the part of bookkeeping nobody enjoys and nobody is uniquely good at. Treatment decisions, sign-off, and advice stay human.
Yes, and that is the best way to start. Running it against a closed month gives you a known answer to compare against before you rely on it for a live one.
It stays in the systems you connected. Access is scoped per tool, logged, and revocable at any time, and Cedric reads the context a task needs and nothing else. Details are in the [privacy policy](/privacy).
Workspaces start with $50 in credits, free, and tasks consume credits in proportion to the work. A monthly reconciliation is a bounded, predictable job; see [pricing](/pricing) for the plan sizes.
Ready when you are
Give the close back its week.
Install Cedric where your finance team talks and start with one reconciliation. It never moves money without your yes.
$50 in free credits, no card needed.