From change to decision, with the evidence
An owner does not need another dashboard. There are enough charts — what is missing is the chain: what changed, why it changed, what follows if we do nothing, and which three things I can do today. NG One is built around that chain. The Executive Cockpit leads from a figure to its cause and to a decision with evidence, the Atlas shows how the whole company runs as several parallel value chains, and the automation KPI answers the question nobody asks out loud: what did this system actually do instead of my people. Underneath that chain sits the whole system — nine business spaces, some 60 technical domains beneath them, eight end-to-end processes — because a cockpit that does not sit on the ledger is a slide.
- Executive Cockpit
- Atlas: map of the company
- Three decisions with what-if
- What the system did on its own
- One source of truth
- Every figure to its document
What costs an owner the most
Not the price of the system. The time between the moment something changed and the moment you found out.
Three reports about the same month, three different numbers
Sales bring their spreadsheet, finance theirs, production a third — and all three are correct in their own world. The meeting is then not about the business but about reconciling sources, and that meeting repeats every month. By the time everyone agrees an hour later on which number is right, the time for a decision has been spent arguing about data.
I know the margin fell, I do not know why
The number is visible; the cause is guessed. Whether it fell because of discounts, purchase prices, product mix, exchange rates or one customer is established by someone digging through spreadsheets for two days. By the time it is known, the next month has passed too, so the same mistake was paid for twice. An answer that arrives on the fifteenth is not an answer but an autopsy.
We paid for a system and people still retype
Nowhere does it say what the system did and what the people did. If there is no number stating how many documents were posted automatically and how much of the statement was matched without a person, then “automation” is a word from the proposal rather than a state of affairs in the company. The investment is measured by the fact that invoices went out before as well — just from a different screen.
I do not understand how my company works inside that system
The menu has fifty entries, each in someone else's language, and the process map exists only in the heads of three people — and leaves with them when they go. The owner therefore does not enter the system but asks someone to pull a report, and so becomes a bottleneck himself. A system that is not understood is not used, no matter how good it is.
Decisions are made on instinct because the data is late
An experienced owner's instinct is better than people admit — but it is better because it is fed by data, not instead of it. When data arrives six weeks late, instinct is the only tool left, and then the stakes are large: price, customer, credit, hiring. Data that arrives after the decision is not data but cost.
How NG One answers
With three gestures built for exactly this chair — and one number that must be yours rather than ours.
Executive Cockpit: change → cause → consequence → three decisions
Not a grid of charts but a chain. The cockpit starts from what changed, breaks the cause down through a waterfall (price, volume, mix, discount, purchase cost, FX), shows the consequence of doing nothing, and offers three decisions with what-if analysis — each with evidence and drill-down to the document. The waterfall is not drawn from a prepared cube but computed from ledger lines, so every ingredient of the breakdown opens down to the document beneath it.
Atlas: a map of the company, not a grid of icons
The Atlas shows the company as several PARALLEL value chains — revenue, purchasing, operations, finance — because a company is not one linear process: purchasing does not always follow sales, and service is an independent flow. Chains cross at shared nodes (stock, invoice). It is not a static picture: the nodes carry live state — how many quotations await a reply, how many tasks are late, how much was processed automatically. The Atlas spans all nine business spaces and the roughly 60 technical domains beneath them — the only map in the system on which an owner sees the whole company at once.
Automation KPI: what the system did on its own
The only number that honestly measures whether the system paid off: how many documents were posted automatically, how much of the bank statement was matched without a person, how many drafts were prepared and how much time that freed. The number comes from YOUR tenant and your data — this site carries no percentage claiming your result, because that percentage does not exist until the system runs at your company. Behind the number sit mechanisms that open up themselves: OCR with per-field-type confidence thresholds, statement matching rules, and drafts a person confirms — each of them leaves a trace in the AI journal.
An AI daily brief and answers with evidence
Instead of generic charts, the morning starts with a sentence that means something: how many decisions are waiting, which collections are at risk, what is unreconciled, and what the system finished on its own since your last login. Every question is asked against the current screen, and every AI conclusion carries an explanation, source documents and drill-down to the data. No evidence, no conclusion — that is a rule, not an option. The copilot works inside your permissions and your tenant: it does not see what you cannot see, and every question and every proposal enters the AI journal.
One source of truth, by architecture rather than by agreement
Three numbers about the same month are not solved by a meeting but by the second source not existing. The dimension is part of the ledger line from the first posting, so a report by plant, project or channel leaves the same ledger as the trial balance and cannot diverge from it. That is a foundation decision: a dimension that was not there at the first posting cannot be added later without reposting history, which is why NG One carries it from the first line.
Every figure opens down to its document
A KPI that cannot be opened is an opinion in a large font. In NG One every figure on the cockpit, in a report and in an AI answer leads to the line and the document it came from — and the posting also references the rule version that produced it, so the explanation is reproducible too. The owner therefore does not have to take the system's word for it, which is the only kind of trust that lasts.
From signal to decision
The chain the Executive Cockpit is built around. Every step has its own screen, and not one of them asks you to take anything on trust.
- Step 1
Signal
The system reports that something changed — while the change is still fresh and the decision still exists.
- KPI movement against plan and against the prior period
- Exceptions and delays on the Atlas nodes
- AI daily brief on the My Work screen
- Collections at risk and unreconciled documents
- Step 2
Cause
The figure is broken into ingredients. Not “margin fell 3%”, but how much of it is price, how much mix, and how much one customer.
- Waterfall breakdown: price, volume, mix, discount, FX, cost
- Impact by dimension: plant, project, channel, customer
- Comparison against plan and against the same period last year
- Separating one-off effects from the trend
- Step 3
Evidence
Every claim opens. This is the step that separates an insight from an opinion.
- Drill-down to the line and to the source document
- The rule version the posting was created under
- Source documents attached to every AI conclusion
- The same answer for everyone who asks — auditors included
- Step 4
Consequence
What happens if I do nothing. The most commonly skipped step, and the only one that makes a decision urgent or unnecessary.
- Projection from open items and contractual terms
- Impact on liquidity and on due dates
- Cash-in and per-partner collection-risk prediction
- Detection of deviation from the usual pattern
- Step 5
Decision
Three options, not thirty. Each with an estimated outcome and the ability to try it before confirming it.
- Three proposed actions with what-if analysis
- The proposal is a DRAFT — confirm, edit and reject are equal
- Approval against limits through the workflow kernel
- A task arises from the decision, with an owner and a deadline
- Step 6
Trace
A decision that is not measured is an opinion. Here it records who decided, on what basis, and how it turned out.
- Decision, reasoning and outcome in the immutable journal
- Automation KPI: what the system did on its own
- Comparison of the expected and the actual outcome
- Delegation and cover without losing accountability
The spaces that carry this solution
A solution is neither a separate product nor a separate licence. It is the same system seen from one angle, and these business spaces carry most of the work this page describes.
Insights, automation and AI
The command center for decisions — evidence behind every insight, a trail behind every automation.
30 capabilitiesMy work
The personalized home screen: decisions and exceptions, tasks and approvals, an AI daily brief, and what NG One finished on its own.
30 capabilities
Questions about this solution
Scope, boundaries, and the rules this entry point works by.
What do I see on day one, and what only once the system has history?
On day one everything deterministic works: positions, ageing, exceptions, tasks awaiting a decision, unreconciled documents, and the automation KPI measuring what the system did on its own. None of that needs history — it only needs the data to be in one system. Predictions are a different kind of thing: cash-in projection, collection risk per partner and anomaly detection are computed from your own patterns, so they sharpen the longer the system watches your business — in the first month they give a rough estimate, after a year they give an estimate you can plan against. We say so in advance because the difference between “I see what is” and “I see what will be” is the one most often erased in proposals.
Where do the “three decisions” come from — does the AI decide instead of me?
No. The cockpit proposes, you decide, and that is not a legal disclaimer but the way the system is built. Every proposal is a DRAFT with the reasoning behind it and the evidence it rests on — confirming, editing and rejecting are equal outcomes and all three stay in the trace. What-if exists so you can see the outcome before you choose it. The reason for three rather than thirty options is practical: a list of thirty proposals is not help but the work handed back to you. If someone promises you an ERP that makes business decisions on its own, ask them who answers when the decision is wrong.
What is the Atlas and why is it not just another dashboard?
The Atlas is a map of the company, not a report. It shows the business as several parallel value chains — revenue, purchasing, operations, finance — because a company is not one linear process: purchasing does not always follow sales, production does not always follow stock, and service is an independent flow. The chains cross where they genuinely cross: at stock and at the invoice. It is not a static picture: the nodes carry live state — how many quotations await a reply, how many tasks are late, how much was processed automatically — and clicking a node leads into the work, not into a legend. For an owner it is worth more than the menu: the menu shows where people work, the Atlas shows how the company functions.
How do I measure whether the ERP paid off?
With the automation KPI, and it is the only number we consider honest in this conversation. It measures how many documents were posted automatically, how much of the bank statement was matched without a person, how many drafts were prepared and how much time that freed — all from your tenant and your data. You will notice this site carries no percentage claiming your result: such a percentage does not exist until the system runs at your company, and a number copied from someone else's case is marketing, not evidence. What we can show beforehand is a demo tenant with twelve months of data — and we will say that it is a demo, not your result.
What in NG One cannot be added later?
The question worth putting to every vendor, because the answer separates a product from a facade. Five things are part of the foundation in NG One and therefore work from the first line, instead of being bolted on once they start to hurt. Dimensional posting: the dimension is part of the ledger line, not a label attached afterwards. Effective-dated statutory parameters: rates, contributions and thresholds are held with their period of validity, so a past calculation reproduces under the rule that applied then. A tenant model with database-level isolation, ready for agency mode where one accountant runs several companies. A domain graph with integrations kept separate, so SEF or the bank never enter posting through a back door. And clean-core extensions with versioned configuration, so fitting the system to your business does not block its upgrades. None of the five can be added to a system that lacks them without reposting history — which is why we do not sell them as a module: they are the reason everything else works.
See the cockpit on your own case
Book a review for management. We take one of your questions — why the margin fell, where the cash sits, who is late — and show the whole chain: from the figure, through the breakdown of its cause, to three decisions and to the document under each one.