OS Finance
Company money from the moment someone asks to spend it through to the bank line that confirms it left — and the budget row it lands in. Requests, approvals, reconciliation and cash in one contour.
Companies with several lines of business, where spending passes through different people and decisions get made against a budget. It answers where the money goes and whether there is enough of it for what is already committed.
Questions it answers
- How much of the balance is actually yours to spend
- When the balance goes negative, and because of which payments
- What a queued request will do to its budget line before it is approved
- What you earned, once other people's money is taken back out
Four screens
What the product actually looks like in use.
Four screens, in the order the questions arrive: what is in the account today, where each line came from, what is left of the plan, and whether the month was actually a good one.
How much cash do we have today, and what happens to it?
Sample dataDashboardSample month
updated today at 09:40
- Free cash
- €84,300of €231,500 across accounts
- Reserved for orders
- €147,20018 active requests
- Clients owe, this month
- €62,4004 requests
- We owe partners, this month
- €51,7009 payments
| Upcoming movement | When | Amount | Balance after |
|---|---|---|---|
| Client balance duerequest 214 · Andersson K. | in 3 days | +€19,325 | €103,625 |
| Partner, second instalmentrequest 214 · pass-through | in 10 days | −€12,400 | €91,225 |
| Team payrolloperations · 6 people | in 15 days | −€14,800 | €76,425 |
| Partner commitmentrequest 221 · pass-through | in 30 days | −€21,500 | €54,925 |
| Further commitmentsweeks 5–6 · 3 payments | in 6 weeks | −€63,825 | −€8,900 |
Where did the money come from, and whose was it?
Sample dataStatementSample month · EUR account
248 transactions · 12 need review
| Date | Counterparty | What it is | Amount |
|---|---|---|---|
| the 24th | Northline Logistics FZCOinvoice 111241 | Client paymentRequest 237 | +€54,900 |
| the 21st | Harbourview Lodge50% deposit | Pass-throughRequest 237 | −€30,748 |
| the 19th | Meridian Supply Ltdparts, operations | Operations · Parts | −€1,812 |
| the 13th | Coral Travel Agencyinvoice 213 | Client paymentRequest 213 | +€6,000 |
| the 11th | Transfer between own accountsEUR → USD, for payroll | Not counted | −€11,200 |
| the 8th | Dockside Provisionstrip supplies | Supplies · Groceries | −€486 |
| the 5th | Terminal 257964card settlement | Where is this from? | +€2,910 |
What did we plan to spend, and what has actually gone?
Sample dataBudgetsOperations · sample month
plan €27,600 · spent €24,300 · pending approval €4,100
| Line item | Plan | Spent | Execution |
|---|---|---|---|
| Parts & maintenance | €9,000 | €11,100 | 123% |
| Fuelper-trip rate · 42 trips | €7,400 | €6,850 | 93% |
| Team payroll | €8,200 | €4,100 | 50% |
| Insurance & licences | €1,900 | €1,900 | 100% |
| Consumables | €1,100 | €350 | 32% |
Did we earn money this month, or just move other people's through?
Sample dataCash flowSample month
all accounts and cash, converted to EUR at the month's rate
- Start of month
- €196,400
- In
- +€312,800
- Out
- −€277,700
- End of month
- €231,500
- Accumulated this quarter
- €61,900
| What kind of money | In | Out | Ours |
|---|---|---|---|
| Client payments on requests | €241,300 | — | €241,300 |
| Pass-through to partnersaccommodation, transport, transfers on client requests | — | €186,900 | −€186,900 |
| Supplier refunds | €3,000 | — | €3,000 |
| Company revenue | €57,400 | ||
| Company expensesoperations, back office, supplies | — | €74,100 | −€74,100 |
| Bank and FX charges | — | €6,700 | −€6,700 |
| Partner capital contributioncapital movement, not revenue | €68,500 | — | not revenue |
| Shareholder loan repayment | — | €10,000 | not an expense |
AI Agents
Agents that sit on top of the operational data and keep watching.
Not a chat window bolted onto a dashboard — nobody has time to interrogate their own software every morning. The agents run against live operational data and speak up on their own. Two of the three below are running today; the third is honestly marked as roadmap.
AI CFO
Watches the money for the things that only become visible too late: a balance turning negative six weeks out, a budget line drifting past its ceiling, a month whose headline balance rose while the company actually lost money. It raises them while a payment can still be moved, rather than reporting them once the month has closed.
Business Analyst
Answers why. It compares this month against the history of this business — the same department, the same line item, the same month last year — and separates a genuine change from the ordinary spread. A figure that looks alarming but sits inside your normal variance gets said out loud as such, instead of triggering an alarm.
Market Research
Where you sit against the market around you — prices moving in your area, how the season is turning, what comparable operators are charging. This one is not built. It needs external data we do not currently hold, and buying or scraping it well is a project in its own right rather than a switch we can turn on.
What they learn on
Every budget, request and reconciled transaction you have put through it is the base the agents learn to understand your business on — not the market in general.
That is a deliberate limit and the reason the first two roles work at all. A model reasoning about operators in the abstract can tell you what is usually true. One reading your own accumulated history can tell you that this route, on this boat, in this month, is not behaving the way it did the last eleven times — which is the only version of the observation you can act on.
It also means the agents get more useful the longer the system runs, and that they are close to useless in week one. We would rather say that than sell a cold start as intelligence.
The daily digest, with reasons
A digest of figures alone is a number you still have to go and explain to yourself. Every line carries why it moved, so the morning read is a decision rather than a research task.
What an alert looks like
Every agent message carries the rows that produced it. A claim you cannot trace back to its source is a claim we are not willing to put in front of you — and it means you can overrule the agent by checking its arithmetic rather than by trusting your gut against it.
Delivery is wherever the team already is — Telegram for the people on the boat or holding the cash, email or the web app for whoever signs things off.
The closing balance rose to €231,500 this month, but the company did not earn money. Of €312,800 in, €186,900 was client money passed straight to partners and €68,500 was a capital contribution — neither is revenue. Company revenue was €57,400 against €80,800 of running cost, so the month is €23,400 down. Worth knowing before the balance is read as a good quarter.
Scope
What a rollout includes, and what this is not.
Included in the rollout
- A separate installation on your own domain, white-labelled in your branding
- One named engineer assigned to the rollout, and still there after go-live
- Dictionaries configured: departments, expense articles, accounts, counterparties, staff
- Telegram bots in the language your team actually speaks
- Historic statements loaded and first-pass article mapping done with you
- Training by role — accountant, department head, owner
- Support terms set per contract, agreed in writing before you start
Not what this is
- Not accounting software — no journal entries, no tax filing, no payroll
- Not a CRM — it does not run sales or hold your customer base
- No per-charter cost build-up — that is calculated in OS Fleet and read from there
Your domain, your branding
A separate installation, white-labelled. Not a tenant on ours.
A named engineer
One person who knows your rollout, from configuration through to the months after go-live.
Next.js and PostgreSQL
Ordinary, well-documented infrastructure. Nothing here is exotic enough to strand you.
Integrations
CRM
Client requests, amounts, payment status, service mix, cost and the partners who delivered.
Accounting
Export of operations for whoever keeps the statutory books.
OS Fleet
Costs and income per vessel, and the work volumes variable lines recalculate on.
An exchange is possible wherever the other system exposes its data through an interface and holds the fields we need. Scope is settled after we look at the system you actually run, and quoted separately. Configuring the far side — fields, dictionaries, access — sits with you or your own contractor.
Request demo
See OS Finance against your own numbers.
Tell us what you run and how you run it now. We will walk the three screens above with your data shape in mind, and say honestly if this is the wrong tool for it.