You are about to hire a Head of Finance you do not need yet
It is the standard move, and for most founders it is the wrong one. The job that actually needs doing in the first two years is not a person sitting in a seat. It is a finance function that exists, and that you can run without becoming a finance person yourself.
- Bookkeeping that is three weeks behind, so every decision is made on a feeling.
- A forecast that is always out of date. It lives in one spreadsheet, one person understands it, and it is never properly rebased on what actually happened, so the runway date it shows is a date nobody in the room quite believes. The number you most need a grip on is the one you trust least.
- No way to see what a hire or a big spend does to that date before you commit to it, so the decision gets made on instinct and the consequence shows up a quarter later.
- Investor reporting rebuilt by hand every quarter, late, and never quite the same twice.
- Two or three entities that should consolidate and do not, so nobody can say what the group made.
There is usually a second worry underneath that one, and founders rarely say it out loud: if I hand this over, am I still the person who understands my own company, and can I still stand behind the numbers I sign? It is a fair question. A finance function you cannot see into is worse than a late one, because you find out at the wrong moment.
A sprint that leaves a finance function behind
Sixty days, adaptable to ninety when the group is more complicated. At the end you have a finance function that runs on a cadence you set, and documentation good enough that the person you eventually hire inherits something instead of starting over.
The sprint is CHF 15,000, fixed and agreed in writing before anything starts. Four things make up the engagement, and they are the same four in the contract. The first three sit inside the setup fee. The fourth is the monthly that runs afterwards:
- Setup. Sixty days of build, ninety where the group starts from nothing. During those sixty days I take over the finance function myself. That is how the work gets captured: I do it, and as I do it I build it into the system, with my own oversight the second time each item comes round. You are not describing your process to a developer, you are handing it to the person who will run it and then automate it. CHF 15,000 fixed for the sixty-day sprint, and where it runs to ninety the price is agreed and written down on the second call, before anything starts.
- Training. Three sessions of two hours for your team, inside the setup fee, so the people who use it every month were taught it by the person who built it.
- Check-ins. Also inside the setup fee. Fortnightly for the first three months after go-live, monthly for three more, then on request. A problem can be reported at any time and is fixed within 36 hours.
- The run. Priced separately, on top of the setup fee: CHF 3,000 to CHF 4,000 per month for the system, the reporting cycle and continued access to me.
If the acceptance test we write together does not pass, you get the kick-off half back, CHF 7,500 on the sixty-day sprint, and the second half is never invoiced, so the sprint costs you nothing.
How it is paid
- Half at kick-off, when the scope of work is signed, so that both of us have committed to the same written thing before any work starts.
- Half at handover, and only when it passes the acceptance test we wrote together. Not on a date, not on an invoice, on the test.
- The monthly starts after that, once the function is deployed, accepted and running on its own with me still overseeing it. Not while it is being built, and not while it still needs me to hold it up.
What the monthly actually buys
- It keeps getting better. Every correction you make becomes a rule, so a thing you have fixed once does not come back next month. That is the whole reason there is a monthly fee rather than a licence.
- Everything is checked before you see it. Key figures are re-derived by a second, independent route, and where it matters by a different model as well, so a wrong number has to survive two attempts to catch it rather than one.
- A log of everything, kept. Every step, every source, every change is recorded and stays auditable, so any figure can be walked backwards months later, by you or by a diligence team.
- And me. Fortnightly for the first three months after go-live, monthly for three more, then on request, with any problem fixed within 36 hours.
On the sixty-day sprint the first year is CHF 45,000 to CHF 55,000 in total. CHF 15,000 for the sprint, which covers sixty days, then CHF 3,000 to CHF 4,000 a month for the ten months after it. From the second year it is CHF 36,000 to CHF 48,000 a year, twelve months of the run, and at the end of the sprint the systems exist either way. Put that next to the offer letter you were about to sign: a CFO on your payroll is around 150K a year, cost to the company with employer social charges in it.
The books, current and trusted
Bookkeeping caught up, carried by us and kept inside the system, which is where the three-week delay stops being a fact of life. Multi-entity where it applies, consolidated. Month end that closes without a fire drill, and a close that refuses to pass a figure it cannot reconcile rather than carrying it quietly forward.
A model that survives being pulled on
Operations tied to cash, not a revenue curve with a margin under it. Rebased on the actuals every month, so the runway date moves when reality does. Every number traceable back to the entry it came from, and built so an investor can push on it in the room and it holds.
Reporting you approve, not reporting you assemble
The board pack and the investor update arrive drafted from the books, in the same shape every quarter, on time. You read it, change what you want changed, and release it. What you change is learned, so the next one arrives closer to how you would have written it.
The system is modular, and the sprint is not the only way in. A module can be ordered on its own, or two or three of them in a sequence, if that suits you better than the whole function at once. Starting with one is often the sensible move: you find out how I work and how the system I set up actually behaves, on one part of your finance function rather than all of it. When you are satisfied, you take the next module. If you are not, you carry on exactly as you do today, and you have lost a module rather than a year.
- A first call. Thirty minutes, free. We work out whether this is your problem at all. Sometimes the answer is that you do need the hire, and I will say so.
- Scope on the second call, against what we found, with the sequence and what it costs written down before anything starts.
- The sprint. Sixty days, working sessions rather than a report at the end. You see it being built, and it takes on how your company actually works as we go.
- Handover. Documented, and the reporting cycle runs on its own with me still watching it.
The whole finance function, not the parts that demo well
This is the list rather than a summary of the list, because a founder deciding between this and a hire is comparing a job description, and a job description is specific. Everything listed here is inside the sprint for a single Swiss company. What sits outside it is named further down the page, in its own words.
Books, close and execution
- Bank data pulled where there is a feed and read off the statements where there is not, because several Swiss banks still have no usable one.
- Transactions booked and classified on your own history. Where it is unsure it asks, and it remembers your answer rather than asking again next month.
- A close finished by working day three: accruals, prepayments, revenue recognition, inventory where you hold it, bank reconciliations, and an exception queue rather than a silent guess.
- The execution work as well, not only the analysis. Payment files for your bank created automatically from the approved payables rather than keyed in by hand, plus payroll and social insurance, payables, cash overviews and cash consolidation across accounts and currencies. The nitty-gritty of a finance department is included, not referred elsewhere.
- Consolidation across subsidiaries, which a standard bookkeeping tool does not do at all and a Treuhand does with a delay you feel. Intercompany eliminated, one group view, monthly.
- Money never moves by itself. Payments are prepared; you approve and release every one.
Steering, forecasting and reporting
- Variance analysis, both of them. Actuals against the approved budget, and actuals against the last rebased forecast. That pair is what a board asks for first and what you steer the company on: the budget says whether you are on the plan you sold, the rebase says whether you are on the plan you now believe.
- The model rebased on real actuals every month, so the runway date is current rather than a memory of the last raise.
- Scenario analysis. Ask in plain language what two sales hires in October do, or a delayed tranche, or a price change, and get the effect on runway, burn and payback on your own numbers, before you commit. Decisions on modelled outcomes rather than on feel.
- Board pack and investor update drafted from the books in minutes, edited by chat, filed automatically. With the feedback loop closed: what the board asked last time is carried forward and answered in the next one.
- A data room kept continuously current, so diligence starts on the day you are asked rather than three weeks later.
- A watchdog on overspend, budget overruns and anomalies in the daily numbers, which catches the quiet failure earlier than a person reading a monthly report would.
Talk to your finances
The nitty-gritty is done and out of the way: the compiling, the comparing, the updating, the checking, every month, without anybody chasing it. On top of that it analyses, and it knows your current figures by heart. So you can ask it anything about your own company's finances, and test an idea against them before you commit to it. Chat with your automated CFO. It knows your numbers, your KPIs and your strategy inside out, and it is available at any hour to help you run and build the business, not only to report on it.
Ask in plain language and get the answer with the figures behind it and the source of each one, as a message in whatever you already use, Slack or email, or in the system itself. Alerts arrive the same way. Where it cannot ground an answer it says so instead of producing one.
Always calculated, never guessed. The AI moves, classifies and explains. It does not do the arithmetic. Every figure comes out of a deterministic formula in the spreadsheet, traceable to the ledger entry underneath it, which is why a number can be re-derived by someone who was not there when it was built.
And it is fast, which is the part that changes behaviour
The close lands on working day three rather than three weeks later, and between closes the numbers are close to real time. The honest limit is Swiss banking rather than us: where a bank has no feed, the last mile is a file, and we will tell you which of your accounts are which before you buy anything.
Speed is not a vanity metric here. A number that arrives three weeks late is a report. A number that arrives on day three is still a decision you can make.
You have a CFO from day one, not at the end of it
The sprint is not a build that is dropped on you at the end with nobody minding the shop in between. From the first week I run the finance function myself while it is being built, so the board pack goes out, the payments get prepared and the questions get answered from day one. Every task the system is going to take over crosses my desk first and is handled by me. That is what makes it real: the work is captured while a finance person is actually doing it, not described to a developer afterwards. It is also why what you end up with can be handed to a person as easily as to the system, because it was written down as work, not as code.
A forward deployed CFO
I work inside the company rather than reporting on it from outside. The way a forward deployed engineer sits with the customer until the software fits the work, except that the job here is the finance function.
The workflows get captured while they run
Every procedure that touches finance gets mapped as it happens, which is the only time it is accurate. That capture is what the system is built from, and it is also the documentation your eventual hire inherits.
Modelling is the part I am best at
We can work with the budget and the model you already have, improve them first and then make the better version the basis of the system, or build a new one with you during the sprint. Which of the three is right is usually obvious within a week.
The finance function on a screen, before you commit to anything
This is the working demo, and it opens inside this page. It runs on two invented companies, Alpina Bio AG in Basel and Meridian Robotics GmbH in Munich, because no client's numbers belong on a public website. The companies are fictional. The mechanics are the real ones: one calculation engine, typed fixtures, nothing in the interface hardcoded.
It opens on the monthly close, which is where the trust is either earned or lost. The left-hand column walks the rest: the rebase, the scenarios, the board pack, the data room. Every figure on every row carries its own currency code, which is the discipline that stops a CHF number being read as a USD one. Prefer a full window? Open the demo in a new tab.
Or watch it walked through, out loud
Four minutes and fifty-three seconds, on the same invented companies: the close and the bank and FX gate, the rebase that follows it, a scenario priced while the decision is still open, and what the run-out date does when the answer lands. Nothing is requested from Loom until you press play.
Prefer a full window, or want the chapter list and the transcript beside it? Watch it on Loom in a new tab.
And it is adapted to you. None of what you see here is fixed. The screens, the reports, the cadence and the language they are written in are shaped around how your company actually runs and what you want to look at, during the sprint, with you in the room. If the board pack in your head looks nothing like the one in the demo, that is the version you get.
The decision layer. Somebody else already sells the compliance layer.
Bookkeeping, VAT returns and the annual accounts are a solved, industrialised and genuinely cheap market in Switzerland, and it is not the market we are in. What almost nobody sells a founder is the layer above it: the number you steer on, the reason it moved, and what happens to it if you make the decision you are thinking about.
What we do
Forecast and rolling model, the monthly close and the rebase that follows it, scenario testing, runway, variance against budget and against the last rebase, board and investor reporting, the data room, and the execution work underneath all of it. The bookkeeping comes inside the system, and that is deliberate. Bookkeeping that sits outside is where the delay comes from: you wait for someone else's month before you can look at your own. Inside, the close and the figures you steer on are the same movement, so the numbers are current instead of arriving three weeks late. We carry the books and the VAT returns ourselves, exactly as we do for our own company and for another Swiss firm.
What we do not do
Not your annual accounts and not your audit: your Treuhand stays, and the year end stays with them or with your auditor, who should be doing it. Not legal or tax opinions. Not a cheap self-serve tier, because at that price the work stops being senior and the thing you bought stops being a CFO.
Two more limits, said now rather than in an annex later. The system prepares and a person releases, so nothing goes out in your name without you seeing it, and where it cannot ground a figure in something traceable it stops and asks rather than filling the gap. And it is not live to the minute: the promise we write into your acceptance test is a close finished by working day three, once the bank and Treuhand inputs are in, because not every Swiss bank offers a usable feed yet and the last mile is still a file.
An entrepreneur who has built the function, not an adviser who has seen one
- Companies built and carried personally, one recently sold, and still on the board of another. That is the part that decides how this goes: I have made the decisions you are making, with my own money behind them.
- The finance function built from nothing three times rather than inherited. Most recently as the sole finance leader at KetoSwiss, a venture-capital funded Swiss life-sciences startup, October 2023 to April 2026, leading a completed CHF 3.5m round end to end.
- Payroll, social insurance, contracts and hiring came with it, because in a small company that is the same job.
- Sectors I have actually worked in: physical goods and subscription businesses, capital-intensive assets and real estate, R&D-heavy companies, international trade and food processing. Where your sector is on that list, the first conversation starts a long way in.
- Not a software engineer and no Fachausweis, and you should know that now rather than later. What I am is an experienced entrepreneur with twenty years of finance under it, who knows how to get things done and how to make a workflow efficient. What is on offer is the finance function and the systems under it.
I am an entrepreneur before I am a finance person. I have built companies and carried them, built the teams inside them, and built the processes from nothing rather than inheriting them. Then I signed for the numbers those processes produced. What is unusual is not any one of those, it is the combination: an owner-operator, twenty years of finance under it, and a working, current command of agentic AI. All of it points at one thing, which is more value created in your company.
- An entrepreneur first. Companies built and carried personally, one recently sold, and still on the board of another.
- Twenty years in finance, nine as a CFO and owner-operator, with the finance function built from nothing three times rather than inherited.
- The sole finance leader at KetoSwiss, a venture-capital funded Swiss life-sciences startup, leading a completed CHF 3.5m financing round end to end.
- Investment and forecasting at a Swiss asset manager, with CHF 6.5m saved on a co-led CHF 200m+ development.
Built inside physical goods and subscription businesses, capital-intensive assets and real estate, R&D-heavy companies, international trade and food processing. We build custom, for one company at a time, in the sectors we actually know.
What I am not: a software engineer, and I do not hold a Fachausweis. I am experienced with AI and with corporate finance. If a Fachausweis is a must-have for you, I am not the right person, and I would rather you knew that now than after a call.
Questions we get asked
Is this instead of hiring a CFO or a Head of Finance?
It is instead of hiring one yet. The sprint builds the finance function first, so that when you do hire, the person inherits working books, a live model and a reporting cycle that runs, rather than spending their first six months building all three. Sometimes the honest answer on the first call is that you should hire now. If that is your situation I will say so and tell you what to look for.
What does it cost?
The sixty day sprint is CHF 15,000, paid half at kick-off when the scope of work is signed and half at handover, and the second half only once the acceptance test we wrote together has passed. Running it afterwards is CHF 3,000 to CHF 4,000 per month, which covers the system, the reporting cycle and continued access to me. The setup fee includes three training sessions of two hours for your team. Where a group starts from nothing the sprint runs to ninety days; the price is agreed and written down on the second call, before anything starts.
What is the guarantee?
Before the sprint starts we write down what "working" means as a concrete acceptance test: the specific things the finance function must do, in writing, agreed by both of us. If it does not pass, you get the kick-off half back and the second half is never invoiced, so the sprint costs you nothing. An unbounded guarantee is worth nothing, so we bound it in the contract instead.
If a system is doing the work, what is left for me?
The deciding, which was always the part worth your time. The system is given the context of your company, how it works and what it values, and it gets more like you over time rather than less, because every correction you make is how it learns. It prepares and you release. Where it cannot ground a number in something traceable it stops and asks instead of guessing. It does not replace you. It represents you, so that you are not in every loop.
Do you replace our Treuhand?
No. Your Treuhand stays, and the annual accounts and the audit stay with them: that is their work and they should be doing it. The bookkeeping is different, and we take that inside. Bookkeeping that sits outside the system is where the delay comes from, because you are waiting on somebody else's month before you can look at your own. Inside, the close and the figures you steer on are one movement, and we carry the books and the VAT returns ourselves, as we already do for our own company and for another Swiss firm. What changes for you is that the numbers arrive current and in a shape you can decide on, instead of three weeks late as a PDF.
How is this different from the digital bookkeeping tools?
Different layer. Those sell the compliance layer, bookkeeping, VAT and year end, at a price that only works with thousands of clients, and they are good value for exactly that. None of it tells you when you run out of money, what moved the number since last month, or what a hire in October does to the runway. That is the layer we build, and a company can sensibly buy both. If a tool is the only finance you have today, keep it and add this on top.
Can we see it before we buy it?
Yes. The working demo is on this page and opens in place, on two invented companies, with no login and nothing to install. On the call we walk through it against your situation rather than the invented one, which is a more useful half hour than any deck.
Which sectors do you work with?
Sector-agnostic on purpose. The machine that turns money into output looks the same whether it is making furniture, software or protein. What matters far more than your sector is whether your cost base has outgrown what you can hold in your head, and whether finance is somebody's actual job yet.
What happens after the sprint ends?
The function runs, and you work with it however suits you. It sends you the alerts, you ask it questions the way you would ask a CFO, and the reporting cycle arrives without anyone assembling it. The system itself stays ours and we keep it running, current and learning your company: that is what the monthly fee is for. Your data, your documentation and your numbers are yours throughout. And for anything the system cannot help you with, I am still there: fortnightly check-ins for the first three months after go-live, monthly for three more, then on request, with any problem fixed within 36 hours.
How do we pay, and when?
Half of the setup fee when the scope of work is signed, half at handover and only once it passes the acceptance test we wrote together. The monthly fee starts after that, when the function is deployed, accepted and running on its own with me still overseeing it. You are never paying a monthly for something that is still being built.
Can you consolidate our subsidiaries?
Yes, and it is one of the clearer reasons to do this at all. A standard bookkeeping tool does not consolidate; a Treuhand will, with a delay you feel every time you need the group picture. Here the group view is produced monthly with the rest of the close, intercompany eliminated, in the same shape every time. Multi-currency comes with it, and every figure carries its own currency code rather than inheriting one from a heading.
Can we ask it questions ourselves?
Yes, and that is how it is meant to be used. Ask in plain language, in Slack or by email or in the system, and you get an answer with the figures behind it and the source of each one. The important part is what happens underneath: the arithmetic is done by deterministic formulas in the spreadsheet and traced back to the ledger, never produced by a language model. It moves and explains the numbers. It does not invent them, and where it cannot ground an answer it says so.
Can you show me a reference?
References on request. Ask on the call and you will be given names and numbers to ring. Client names and figures are not published without written permission, which is the same discipline that would apply to yours.
Something not answered here? Ask it on the call, or ring +41 44 798 55 85.
Thirty minutes to find out if this is your problem
No preparation and no pitch deck. If a permanent hire is genuinely the right answer for you, I will tell you what to look for.
Speak to Oliver to find out moreOr ring +41 44 798 55 85. References on request.