FINPLAI is controlling and planning with AI at the core. It runs on your infrastructure, your data never leaves the company, and a manager's question is answered in five minutes instead of two days.
The Q3 executive summary is ready.
Drafted over the weekend, in the background. Nobody wrote it.
Why is margin falling in Slovakia?
Butter from supplier B: +24% above the contracted price.
The input price rise behind the largest share of the margin drop. Suggested action: reopen the contract. Caught on the day it happened, not at month-end close.
What if we raise sandwich prices by 4%?
Full-year forecast recalculated, price approved.
The decision rests on today’s data, not on a report that is two days old.
The new customer’s request has been priced.
The costing ran by itself the moment sales entered volumes and dates.
What do we actually make on this deal?
Logistics to that region: +38% above the usual rate.
The costing assumes a one-off delivery. On a regular route the item halves.
What if we go 3% lower against a volume commitment?
Quote sent, conditional on a minimum volume.
Lower price, higher total contribution. Worked out before sales could call back.
The 2027 sales plan is in from every division.
Bottom-up from six managers, aggregated and checked automatically.
Why doesn’t the plan match our production capacity?
Planned volume exceeds line 2 capacity by 16% in September.
Found during aggregation, not when production started missing dates. Suggested action: move volume to line 3.
What if we move a fifth of the volume to line 3?
Volume moved, the sales plan stays untouched.
Sales did not have to cut the target. Only the route to producing it changed.
Next quarter’s cash flow has been recalculated.
It ran overnight, right after the bank statements landed.
Why is our working capital climbing?
Customer C pays in 62 days on average, against 30 contracted.
On its own that ties up 8.9 M CZK of the increase. Suggested action: stop further deliveries on invoice and move to prepayment.
What if we push the production line investment out by a quarter?
Investment moved to Q1, customer C’s credit limit cut.
Both written into the plan before the meeting ended.
The year-end forecast has been refreshed.
It recalculated on its own as soon as September was closed.
Where are we short of the annual target?
Retail holds volume, but the average price has fallen 6%.
A promotion launched for one month has been running for five months straight. Nobody switched it off, because nobody was watching it.
What if we end the promotion on 1 December?
Promotion ends 1 December, the target is back within reach.
A five-month-old discount found its owner in twenty minutes.
Not because the team is weak, but because its capacity disappears into mechanical work that nobody sees and nobody pays for.
A controller spends dozens of hours a month mechanically consolidating data instead of interpreting it.
An ad-hoc question — “why is margin falling in Slovakia?” — takes two days. By the time the answer lands, the decision has been made without it.
A jump in costs shows up at month-end close. The response comes late and the loss can no longer be recovered.
The annual forecast is refreshed quarterly, but the market moves in weeks.
The controlling team rewrites more data than it interprets. People's capacity goes into mechanical work, not business insight.
A typical mid-market CFO, 2026
Three roles working side by side. None of them touches your data on its own — the AI proposes, a person always writes.
The manager asks in plain language and gets an answer backed by data. No SQL, no waiting on controlling.
Helps build the plan — suggests adjustments, computes what-if impacts, points out inconsistencies.
Works in the background. Watches the data, surfaces deviations and drafts the executive summary before you ask.
You don't have to pick one and keep the rest in Excel on the side. The methods combine, and you can switch between plan versions.
Leadership sets the overall targets, the AI breaks them down by division, customer and period.
Unit managers plan their own numbers, FINPLAI aggregates them and watches for inconsistencies.
A plan driven by business drivers, with the relationships between them checked automatically.
The forecast refreshes continuously, with updates proposed by the AI.
Several plan versions side by side, with impacts computed live.
Every version auditable, with a change log and authorship.
Rollout
Controlling changes constantly. If every new dimension meant raising a ticket with the vendor, the system would go stale as fast as the last one did.
Data model through the UI
Define dimensions, measures and relationships without SQL or XML.
AI tool API without code
Configure AI access through validated, typed calls.
Security roles in the UI
Manage permissions without touching server administration.
Vendor independence
The controller can adjust the system alone, long after rollout ends.
The controller moves from data courier to analyst and advisor. That is the whole business case — the numbers below only confirm it.
The most common objection to AI in finance is not price. It is what happens to the data and who is accountable for the decision. Here are the answers up front.
The AI is an extension, not a dependency. Switch it off and the system keeps running in full — every AI function has a manual alternative. You are not held hostage by the technology.
The AI provider Ollama runs inside your perimeter. No outbound calls to a public cloud — and no compromise on capability.
The AI only reads and proposes. Data is changed by a user, after confirmation. No surprises from the AI, no runaway agent.
The AI has no direct database access. It works through typed, validated calls — no SQL injection, no unbounded queries over your data.
Every question, every answer, every call logged with a trace ID. Anything can be reconstructed after the fact.
The AI respects the user's role — a controller sees different answers from a sales rep. Isolation between tenants is enforced at the API layer.
The NIS2 directive (EU 2022/2555; in the Czech Republic Act No. 264/2025 Coll., in force since November 2025) requires companies to put cybersecurity measures in place. FINPLAI is built to support your compliance — access control, encrypted transport, tenant data separation, audit logging, backup and recovery, on-premise and air-gapped operation. We will supply security documentation for your supplier register. Multi-factor sign-in (MFA / SSO) and ISO 27001 certification are on the roadmap.
Excel is flexible and nobody is taking it away from you. The risk is not that it stagnates — it is the errors in the formulas and the fact that you cannot govern it.
| Criterion | FINPLAI | SAP BPC | Anaplan | Excel + BI |
|---|---|---|---|---|
| AI assistant in the UI | ●native, three pillars, streaming | ○Joule add-on only | ◐CoPlanner in pilot | ○none |
| On-premise with local AI | ●Ollama, data never leaves | ◐on-prem yes, AI no | ○cloud only | ◐on-prem, AI no |
| Time to value (rollout) | ●2–3 weeks pilot | ○6–12 months | ◐3–6 months | ◐immediate, but unstructured |
| 3-year TCO, mid-market | ●low, fixed price possible | ○very high | ○per-user, high | ◐cheap to license, expensive in hiding |
| Auditability, error risk | ●fully audited, AI never writes | ◐depends on the rollout | ○data in the cloud | ○formula errors, unmaintainable |
Most of the company only needs to read. A handful of people plan, and an even smaller group uses predictive AI. The three tiers follow that shape.
Management · team leads · board
Finance managers · controllers · product & sales
CFO · senior controllers · FP&A · BI analysts
We price against your mix of roles
How many people only read, how many plan, and how many need predictive AI. You get a number on the first call, not after the third meeting.
FINPLAI runs alongside and does not touch your existing systems. Fixed scope, fixed price. It will not break your running processes mid-quarter.
One server or VM, Linux + Docker. Scope per the pilot one-pager — nothing beyond it.
Onboarding from your cost-centre catalogue, import of a data sample.
All three AI pillars demonstrated on real numbers, plus user training.
Runs in parallel
Your existing processes continue unchanged. The pilot can be stopped at any point.
Fixed price
Scope and price are agreed up front. No change requests at the end.
Your data, your server
The pilot runs on your side. The data sample never leaves your perimeter.
FINPLAI continues work that started long before anyone talked about AI in controlling. It rests on two foundations: twenty-five years of planning and controlling projects in Czech and Slovak companies, and twenty years of in-house development in business intelligence and enterprise software.
Contributed to the design and rollout of planning and controlling solutions at dozens of major companies in Czechia and Slovakia. Nearly twenty years at Inekon Systems, six of them as chief executive. He knows why most planning projects end up back in Excel — and what a system has to do to avoid it. That knowledge is what we build into FINPLAI, on today’s technology.
Twenty years building management information systems and solutions for data integration and reporting at many Czech companies. He started at 2N Telekomunikace as a developer and left as its head of IT. Since 2013 he has co-owned SOFTIM.CZ with his son Jan Hnízdil — a company continuing a tradition of software development for business intelligence, and, as co-owner of FINPLAI, supplying its software and AI. No outsourcing, and no black box that nobody can change later.
Plans get entered where people hold them in their heads.
A director thinks in company totals, a sales rep in individual customers. The system has to accept both and work out the rest. This decided whether people used a planning system or worked around it.
A new version must not mean starting over.
Copy actuals into the plan with every ratio preserved; use the plan as the basis for the forecast. Without it, controlling spends every cycle retyping numbers it already entered once.
If a recalculation takes hours, nobody runs scenarios.
Recomputing millions of values in minutes is not a technical flourish. It is the difference between planning that leads to a decision and planning that gets postponed.
Thirty minutes, a walkthrough on your data sample, no commitment. You leave either with a pilot date or with a clear reason why not.