● B2B offer — Executive management
Executive cockpit: costed scenarios, every assumption visible
An executive committee decides between scenarios: an investment, an opening, a closure, a change of model. This cockpit consolidates your management data and costs several scenarios, each with its assumptions, its source data and its sensitivity to the key variables. Hosted in France: all of your management data stays with you. The committee weighs it up and decides.
Hosted in France
Executive data protected
GDPR & AI Act: governed deployment
Human oversight
Updated on
Deployed in a few weeks
Executive cockpit · hosted in France
Cost the opening of an additional site in this area.
Scenario costed from your consolidated data: costs, volumes served, effects on the existing flows.
The assumptions are listed one by one, with the source data for each.
The sensitivity to the two most decisive variables is set out.
🔗 Sourced · consolidated data, assumptions listed
The assumptions are listed one by one, with the source data for each.
The sensitivity to the two most decisive variables is set out.
🔗 Sourced · consolidated data, assumptions listed
What decision do you recommend?
I set out the scenarios and what separates them: diverging assumptions, sensitive variables, effects on existing activities.
An executive decision takes in things that are not in the data — strategy, market, appetite for risk. It belongs to the committee.
✎ Support · scenarios presented, the committee's call
An executive decision takes in things that are not in the data — strategy, market, appetite for risk. It belongs to the committee.
✎ Support · scenarios presented, the committee's call
Local inference · no data outside the EU
Cockpit hosted in France
Sovereign by designLocal inference or hosting in France
GDPR & AI Act: governed deploymentTraceability & human oversight
TurnkeyDesigned, installed and operated for you
The executive committee decidesThe agent prepares, never rules
✦ In brief
A Blue Lemon Agent cockpit consolidates your management data and costs several scenarios, each with its assumptions listed one by one, their source data and its sensitivity to the key variables. It does not recommend: an executive decision takes in things absent from the data. Hosted in France, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
100%
hosted in France in the target architecture
0
transfer outside the EU in the target architecture
6
executive uses ready to deploy
0
decision taken without human approval
These figures describe our offer, not results measured at a client. How large the gain is on the number of sources consolidated and scenarios modelled is confirmed by a pilot.
The context
What does an AI cockpit bring to your executive committee?
A scenario whose every assumption is visible can be discussed in committee; an isolated figure invites dispute.
! The issue
An executive decision rests on comparable scenarios and on confidence in their assumptions. Rebuilding those scenarios for every committee takes time and leaves room for doubt about the figures. The cockpit costs them from the consolidated data and shows every assumption with its source.
✓ Our answer
The executive committee compares traceable scenarios and sees what separates them: diverging assumptions, sensitive variables, effects on what already exists. The decision takes in strategy, market and appetite for risk — things that are not in the data and that belong to the committee. Local inference or an isolated resource hosted in France: all of your management data does not leave the company.
The decisive point
All of the company's management data: sovereignty & compliance
An executive cockpit brings together the most complete picture of the company. Here is how that concentration is protected.
Local inference
The agent can run on a machine belonging to your organisation: no management data and no scenario leaves the network.
Hosting in France
Otherwise, a dedicated and isolated resource hosted in France, under French law — your consolidated data and your scenarios: processing and access within the European Union targeted by the architecture.
Reduced extraterritorial exposure
For all of the company's management data, the architecture aims to reduce exposure to the Cloud Act and FISA 702; being located in France or in the European Union does not, on its own, guarantee immunity.
Isolated resource
No pooling: an environment strictly dedicated to your company and its decision models.
Every assumption traced to its source
The assumptions are listed one by one with their source data; encryption, role-based access and logging of the scenarios produced.
AI Act: governed deployment
The agent is strictly in support; no recommendation is presented as a conclusion and no decision is taken; traceability and human oversight from end to end.
What depends on the architecture chosen These points are not general guarantees: they are settled deployment by deployment, in the quotation.
- The applicable location is that of the architecture set out in the quotation and verified before commissioning.
- Local execution is announced only for the configuration explicitly described and accepted in the quotation.
- The applicable isolation depends on the deployment mode set out in the quotation; no dedicated isolation is presumed.
- Roles and permissions are configured and accepted for the identities and systems actually connected.
- The events logged, their content, their retention period and who may access them are defined for the deployment chosen.
For scenarios tied to a corporate finance transaction or a confidential reorganisation, SecNumCloud and reinforced hosting are options depending on your requirements. A single architecture is designed to answer both the GDPR and extraterritorial exposure. Designed for deployment in line with the GDPR and the AI Act, after the processing, roles and context-specific risks have been assessed.
Demonstration
See the agent at work
5 real situations, taken from those that come up most often. Pick one: the exchange unfolds as it would in your organisation.
A scripted demonstration. These exchanges show how the agent behaves — its sources, its refusals, what it leaves to your teams. Nothing is sent from this page, no model is queried here, and the matters named are fictional. That is precisely what we promise your data.
The behaviours shown here — monitoring, automation rules, routing and reminders — are configured with you during deployment, from your tools, your rules and your thresholds.
The architecture points named in these exchanges — location, local execution, isolation, encryption, role-based access, logging — are not a guarantee attached to the demonstration: they are those of the architecture set out in your quotation, and verified before commissioning.
The company in this demonstration
Fictional companyVaubert & Cie — manufacturer and maintainer of commercial refrigeration equipment
- Sector
- Manufacture of commercial refrigeration equipment, sales and maintenance contracts
- Headcount
- 340 staff, including 6 on the executive committee, 2 management accountants and 1 consolidation officer
- Market served
- 1,200 customers — food retail, catering, food-processing industry — France and Benelux
- Order of magnitude
- €78M in revenue, 4 production sites, 780 maintenance contracts, 11 executive committee meetings a year and 14 investment decisions
- Tools in place
- ERP, cost accounting, CRM, maintenance management, HR system and 40 spreadsheet workbooks — the cockpit connects to them, nothing is replaced
- Who decides
- The executive committee arbitrates the scenarios; the finance director settles the assumptions; the supervisory board approves anything above €2M
- Room for improvement
- Consolidating for one board meeting takes 9 working days from two management accountants; one costed scenario takes 4 to 6 days and exists in a single set of assumptions; 3 of last year's 14 investment decisions were postponed for want of figures in time
Vaubert & Cie is not short of data: it is short of time between the moment the figures are ready and the moment the board has to decide. The cockpit runs on local inference on a machine at head office and connects to the ERP, cost accounting, the CRM, maintenance management and the HR system: it consolidates and costs, the board arbitrates. The exchanges below cover one quarter, from the first connection of the sources to the review meeting.
This company, its figures and the exchanges that follow were invented for the demonstration. They illustrate a common situation; they describe no real client.
Executive cockpit · hosted in France
Your six sources have been consolidated since last night, and the file for the meeting on the 14th is ready. What your two management accountants used to do in 9 working days is done in 4 hours — but that is not the most interesting figure in this message.
The gap I found by reconciling the sources, and it was visible in none of them: your margin per site is calculated under two different rules. Two of your sites book warranty maintenance to the contract, the other two book it to production. Over the year the gap is worth 1.4 margin points — €1.1M — and it always falls on the same two sites. This is not an error, it is a convention that was never written down.
What I did with it: the two rules are set out side by side, each with its result, and I did not choose. An accounting convention is the rule that says where a cost is booked: it does not change the company's result, it changes how two sites compare. Settling on the right one belongs to your finance director — he is the one who knows what the board wants to compare.
What changes at the meeting on the 14th: the four sites become comparable, and the ranking of the bottom two reverses. The Reims site, which the board has taken for the weakest performer for three years, moves ahead of Lille as soon as the same rule is applied to both.
The time this shifts: consolidating the data for a board meeting used to take 65% of the preparation time; it takes 7%. The remaining 7% is your accountants' review, and it is that review their signature stands behind.
What I propose: that the finance director settle the convention before the 14th, in one line. I will recalculate the last three financial years under both rules tonight: the board will see a consistent history, not a break in the series halfway through a chart. consolidation_6-sources-3-definitions.pdfWhat is reconciled, and the gaps found
⛓ Sourced · ERP, cost accounting, CRM, maintenance management, HR system, 40 spreadsheet workbooks
The gap I found by reconciling the sources, and it was visible in none of them: your margin per site is calculated under two different rules. Two of your sites book warranty maintenance to the contract, the other two book it to production. Over the year the gap is worth 1.4 margin points — €1.1M — and it always falls on the same two sites. This is not an error, it is a convention that was never written down.
What I did with it: the two rules are set out side by side, each with its result, and I did not choose. An accounting convention is the rule that says where a cost is booked: it does not change the company's result, it changes how two sites compare. Settling on the right one belongs to your finance director — he is the one who knows what the board wants to compare.
What changes at the meeting on the 14th: the four sites become comparable, and the ranking of the bottom two reverses. The Reims site, which the board has taken for the weakest performer for three years, moves ahead of Lille as soon as the same rule is applied to both.
The time this shifts: consolidating the data for a board meeting used to take 65% of the preparation time; it takes 7%. The remaining 7% is your accountants' review, and it is that review their signature stands behind.
What I propose: that the finance director settle the convention before the 14th, in one line. I will recalculate the last three financial years under both rules tonight: the board will see a consistent history, not a break in the series halfway through a chart. consolidation_6-sources-3-definitions.pdfWhat is reconciled, and the gaps found
⛓ Sourced · ERP, cost accounting, CRM, maintenance management, HR system, 40 spreadsheet workbooks
Forty spreadsheet workbooks — that is our curse. Are you replacing them?
No, and that is deliberate: I read them, I reconcile them, and I show you which ones no longer serve. Replacing first means losing what nobody wrote down anywhere else.
What I found in the 40:
· 11 are genuine sources — they carry management rules that exist nowhere in the ERP: seasonality coefficients, travel allowances by zone, renewal rates by contract family. Those I read at every consolidation.
· 17 are copies of an ERP extract with a rework applied. The rework I redo; the workbook no longer serves.
· 12 have not been updated for more than a year. Two of them still feed a monthly committee — and nobody knows it.
The figure that concerns you: your accountants spent 4.5 days a month maintaining these workbooks, of which 2.8 on the 29 that bring nothing. That is half a management accountant, busy copying.
What gets signed, and why: deleting a workbook. I read them read-only and modify none — a workbook that disappears takes with it a rule its author never wrote down anywhere else, and the author has often left. The work itself is done: the 29 that bring nothing are identified one by one, and the 11 management rules held by the useful workbooks are already extracted, named and ready to become cockpit parameters.
What I propose: that the 11 genuine sources be taken up as cockpit parameters, visible and editable by your accountants without going through me. The other 29 stay where they are: you will retire them once you have seen three months of figures without them. 40-workbooks_which-ones-are-a-source.pdf11 genuine sources, 17 copies, 12 abandoned
⛓ Sourced · 40 management-accounting workbooks, access logs, ERP
What I found in the 40:
· 11 are genuine sources — they carry management rules that exist nowhere in the ERP: seasonality coefficients, travel allowances by zone, renewal rates by contract family. Those I read at every consolidation.
· 17 are copies of an ERP extract with a rework applied. The rework I redo; the workbook no longer serves.
· 12 have not been updated for more than a year. Two of them still feed a monthly committee — and nobody knows it.
The figure that concerns you: your accountants spent 4.5 days a month maintaining these workbooks, of which 2.8 on the 29 that bring nothing. That is half a management accountant, busy copying.
What gets signed, and why: deleting a workbook. I read them read-only and modify none — a workbook that disappears takes with it a rule its author never wrote down anywhere else, and the author has often left. The work itself is done: the 29 that bring nothing are identified one by one, and the 11 management rules held by the useful workbooks are already extracted, named and ready to become cockpit parameters.
What I propose: that the 11 genuine sources be taken up as cockpit parameters, visible and editable by your accountants without going through me. The other 29 stay where they are: you will retire them once you have seen three months of figures without them. 40-workbooks_which-ones-are-a-source.pdf11 genuine sources, 17 copies, 12 abandoned
⛓ Sourced · 40 management-accounting workbooks, access logs, ERP
Our overall margin is 12.4% and we never really know where it comes from. Can you answer that in plain language?
Yes, and the answer is uncomfortable: your 12.4% is an average of two businesses that look nothing alike. Equipment sales pay for maintenance, and nobody had put a number on it.
What the reconciliation shows, contract by contract:
· New equipment sales: 17.1% margin, stable over three financial years.
· Maintenance contracts: 4.8% on average, but the average hides everything.
· Of your 780 maintenance contracts, 214 cost more than they bring in: −€1.1M a year, absorbed silently by the equipment margin.
Why those 214 and not the others — the cause is mechanical, not commercial: they all carry a travel allowance set in 2019 and cover sites more than 90 km from one of your branches. The number of call-outs is no higher there than elsewhere; it is the journey that is no longer covered.
What it is worth if you do nothing but renegotiate the allowance: +€1.1M of margin, that is 1.4 points on group result, without losing a single customer if the adjustment follows the index your contracts already provide for. 15 of these contracts come up for renewal this quarter: that is where to start, because that is where renegotiation is already written into the contract.
What I recommend, and what I advise against, in figures: I advise against terminating the 214 — a loss-making maintenance contract often keeps the customer who buys the equipment, and of those 214, 96 have bought new equipment within two years, for €3.4M of orders. I recommend renegotiating the allowance on the 15 up for renewal this quarter, worth €210k on its own with no contractual risk. That is the real calculation, and it is in the attachment, contract by contract.
And here is where I stop, so you know what you are buying: I read your financial flows to understand a margin, I do not hunt for anomalies in the flows themselves — duplicate payments, atypical receipts, breaks in an audit trail. That is the job of a dedicated financial analysis agent, which connects to the same consolidated foundation. I stop at what feeds a board decision.
What I propose next: I cost three renegotiation options — indexation alone, indexation plus zone, or a full redesign of the allowance — each with the number of contracts at risk of leaving. You choose, I prepare the 15 renewal files for the quarter. real-margin_780-maintenance-contracts.pdf214 loss-making contracts, and why
⛓ Sourced · 780 maintenance contracts, call-outs from maintenance management, cost accounting, sales history
What the reconciliation shows, contract by contract:
· New equipment sales: 17.1% margin, stable over three financial years.
· Maintenance contracts: 4.8% on average, but the average hides everything.
· Of your 780 maintenance contracts, 214 cost more than they bring in: −€1.1M a year, absorbed silently by the equipment margin.
Why those 214 and not the others — the cause is mechanical, not commercial: they all carry a travel allowance set in 2019 and cover sites more than 90 km from one of your branches. The number of call-outs is no higher there than elsewhere; it is the journey that is no longer covered.
What it is worth if you do nothing but renegotiate the allowance: +€1.1M of margin, that is 1.4 points on group result, without losing a single customer if the adjustment follows the index your contracts already provide for. 15 of these contracts come up for renewal this quarter: that is where to start, because that is where renegotiation is already written into the contract.
What I recommend, and what I advise against, in figures: I advise against terminating the 214 — a loss-making maintenance contract often keeps the customer who buys the equipment, and of those 214, 96 have bought new equipment within two years, for €3.4M of orders. I recommend renegotiating the allowance on the 15 up for renewal this quarter, worth €210k on its own with no contractual risk. That is the real calculation, and it is in the attachment, contract by contract.
And here is where I stop, so you know what you are buying: I read your financial flows to understand a margin, I do not hunt for anomalies in the flows themselves — duplicate payments, atypical receipts, breaks in an audit trail. That is the job of a dedicated financial analysis agent, which connects to the same consolidated foundation. I stop at what feeds a board decision.
What I propose next: I cost three renegotiation options — indexation alone, indexation plus zone, or a full redesign of the allowance — each with the number of contracts at risk of leaving. You choose, I prepare the 15 renewal files for the quarter. real-margin_780-maintenance-contracts.pdf214 loss-making contracts, and why
⛓ Sourced · 780 maintenance contracts, call-outs from maintenance management, cost accounting, sales history
The board has to decide on opening a site in the Benelux. Cost it.
Three costed scenarios, comparable with one another, each with its effect on the four sites you already have. That effect is what is usually missing, and it is what makes a board change its mind.
· Scenario A — full site in Antwerp: €3.4M of investment, in service at 14 months, break-even in the 31st month. Effect on the existing network: −6% of load on the Reims site, which currently serves the northern Benelux — and Reims then loses 0.9 margin points, because fixed costs still have to be absorbed.
· Scenario B — extension of the Lille site: €1.2M, in service at 5 months, break-even in the 14th month. But capacity caps at +38%, which covers your volume forecast to 2029 and no further.
· Scenario C — sales office with maintenance subcontracted: €0.4M, operational in 3 months, break-even in the 7th month, but 5 margin points less on the contracts served, and the maintenance relationship goes through a third party.
What I underline before the result, because it is what decides: the three scenarios do not rest on the same bet. A bets on volume, B bets on the calendar, C bets on a partner. A board does not arbitrate three figures, it arbitrates three risks — and I named them before costing them.
The time this shifts: costing a scenario used to take 55% of the preparation time for a board meeting; it takes 9%, and you have three instead of one. Of last year's 14 investment decisions, 3 were postponed for want of figures in time.
What I propose: that the board settle the bet first, not the figure. Tell me which of the three risks you accept, and I will tighten the other two scenarios around it. benelux-opening_3-costed-scenarios.pdfInvestment, break-even, the bet behind each
⛓ Sourced · cost accounting, site costs of the 4 plants, load history, volume forecast
· Scenario A — full site in Antwerp: €3.4M of investment, in service at 14 months, break-even in the 31st month. Effect on the existing network: −6% of load on the Reims site, which currently serves the northern Benelux — and Reims then loses 0.9 margin points, because fixed costs still have to be absorbed.
· Scenario B — extension of the Lille site: €1.2M, in service at 5 months, break-even in the 14th month. But capacity caps at +38%, which covers your volume forecast to 2029 and no further.
· Scenario C — sales office with maintenance subcontracted: €0.4M, operational in 3 months, break-even in the 7th month, but 5 margin points less on the contracts served, and the maintenance relationship goes through a third party.
What I underline before the result, because it is what decides: the three scenarios do not rest on the same bet. A bets on volume, B bets on the calendar, C bets on a partner. A board does not arbitrate three figures, it arbitrates three risks — and I named them before costing them.
The time this shifts: costing a scenario used to take 55% of the preparation time for a board meeting; it takes 9%, and you have three instead of one. Of last year's 14 investment decisions, 3 were postponed for want of figures in time.
What I propose: that the board settle the bet first, not the figure. Tell me which of the three risks you accept, and I will tighten the other two scenarios around it. benelux-opening_3-costed-scenarios.pdfInvestment, break-even, the bet behind each
⛓ Sourced · cost accounting, site costs of the 4 plants, load history, volume forecast
The effect on Reims — where does that come from? That is exactly the kind of figure we will be challenged on in the meeting.
From your own deliveries, line by line — and I am giving you what you need to answer the challenge before it comes.
The calculation, in three verifiable elements:
· What Reims currently delivers to the northern Benelux: 2,340 delivery lines over twelve months, that is 6.1% of its load. Source: your delivery notes, 1 July to 30 June.
· What Antwerp would take over: 84% of those lines, the ones whose customer is within 120 km of Antwerp. The remaining 16% are closer to Reims and would stay there.
· What Reims loses in margin: 0.9 points, because its fixed costs do not fall by as much. That is the effect opening plans forget most often — the new site gets costed, not what the old one stops doing.
The comparison that sheds light, and I give it for what it is worth: you have already lived through this, in 2021, when the Lyon site opened. The load taken from Reims was 5.4% against 7% announced, and margin recovered in 11 months. That precedent is in your own figures, not in some sector rule of thumb — it is the only benchmark that commits you.
What it lets you say in the meeting: not « the cockpit says −6% », but « we measured 6.1% of load concerned, 84% of it transferable, and our 2021 precedent came in 1.6 points below forecast ». Same sentence, and it cannot be challenged the same way.
What I propose: I am adding to the file the prudent variant calibrated on the Lyon precedent — 5.4% taken over instead of 6.1%. Break-even for A then moves from the 31st to the 33rd month. If your decision changes between 31 and 33 months, you need to know that now. effects-on-existing-sites_4-plants.pdfWhat each scenario takes away from the current sites
⛓ Sourced · 12 months of delivery notes, fixed costs of the Reims site, the 2021 Lyon opening
The calculation, in three verifiable elements:
· What Reims currently delivers to the northern Benelux: 2,340 delivery lines over twelve months, that is 6.1% of its load. Source: your delivery notes, 1 July to 30 June.
· What Antwerp would take over: 84% of those lines, the ones whose customer is within 120 km of Antwerp. The remaining 16% are closer to Reims and would stay there.
· What Reims loses in margin: 0.9 points, because its fixed costs do not fall by as much. That is the effect opening plans forget most often — the new site gets costed, not what the old one stops doing.
The comparison that sheds light, and I give it for what it is worth: you have already lived through this, in 2021, when the Lyon site opened. The load taken from Reims was 5.4% against 7% announced, and margin recovered in 11 months. That precedent is in your own figures, not in some sector rule of thumb — it is the only benchmark that commits you.
What it lets you say in the meeting: not « the cockpit says −6% », but « we measured 6.1% of load concerned, 84% of it transferable, and our 2021 precedent came in 1.6 points below forecast ». Same sentence, and it cannot be challenged the same way.
What I propose: I am adding to the file the prudent variant calibrated on the Lyon precedent — 5.4% taken over instead of 6.1%. Break-even for A then moves from the 31st to the 33rd month. If your decision changes between 31 and 33 months, you need to know that now. effects-on-existing-sites_4-plants.pdfWhat each scenario takes away from the current sites
⛓ Sourced · 12 months of delivery notes, fixed costs of the Reims site, the 2021 Lyon opening
And if we dropped the whole thing, what do we keep?
The zero scenario is costed too, and it is not neutral: doing nothing costs something, and I would rather that were written down.
Zero scenario — nothing changes:
· The Benelux volume goes on being served from Reims, with an average delivery time of 6.4 days against 2.1 in France.
· Over the last 3 financial years you lost 11 tenders in the Benelux, and 7 of your sales reports name lead time as the reason. That represents €2.9M of revenue not taken — the figure is a total of announced amounts, not a forecast.
· No investment, no new risk, and your borrowing capacity stays intact for the renewal of the Reims production line, planned for 2029.
What I advise against, and I put a number on it: launching A and B together. Your two management accountants would be tracking two ramp-ups at once, and your cash would fall below €1.8M in the 9th month — below the threshold in your banking covenant, that is, the quantified undertaking you gave your bank, the breach of which reopens the negotiation of the credit line. A scenario that wins on paper and reopens your credit line is not a good scenario.
What I propose to the board: four scenarios put to the vote, zero included, each with its bet, its break-even and its effect on the existing network. And the question to settle first, the one that makes the other three readable: up to which month are you willing to wait for the return?
⛓ Sourced · delivery times by zone, sales reports, tenders lost, cash plan
Zero scenario — nothing changes:
· The Benelux volume goes on being served from Reims, with an average delivery time of 6.4 days against 2.1 in France.
· Over the last 3 financial years you lost 11 tenders in the Benelux, and 7 of your sales reports name lead time as the reason. That represents €2.9M of revenue not taken — the figure is a total of announced amounts, not a forecast.
· No investment, no new risk, and your borrowing capacity stays intact for the renewal of the Reims production line, planned for 2029.
What I advise against, and I put a number on it: launching A and B together. Your two management accountants would be tracking two ramp-ups at once, and your cash would fall below €1.8M in the 9th month — below the threshold in your banking covenant, that is, the quantified undertaking you gave your bank, the breach of which reopens the negotiation of the credit line. A scenario that wins on paper and reopens your credit line is not a good scenario.
What I propose to the board: four scenarios put to the vote, zero included, each with its bet, its break-even and its effect on the existing network. And the question to settle first, the one that makes the other three readable: up to which month are you willing to wait for the return?
⛓ Sourced · delivery times by zone, sales reports, tenders lost, cash plan
In the meeting we spend an hour wondering where the figures come from. How do you handle that?
The 34 assumptions behind scenario A are listed one by one, each with its source data, its date and its author. None of them is mine: they come from your data or from your decisions, and I say which.
Three families, and it is the sorting that makes the meeting short:
· 18 measured assumptions — they come out of your data, with the exact period. Example: hourly production cost at Reims, €42.10, cost accounting, closed financial year. They are not up for discussion, they are up for verification.
· 12 assumptions settled by you — an executive decision already taken, dated, with its author. Example: discount rate of 6%, settled by the finance director on 12 March 2026. They can be discussed, but everyone knows who set them.
· 4 assumptions with no source, and those are the ones that deserve the hour of meeting time. I put them at the front of the file rather than in an appendix.
The time this shifts: documenting the assumptions used to take 30% of the preparation time; it takes 4%. And before that it existed only in the head of the accountant who built the spreadsheet — which made it unchallengeable for all the wrong reasons.
The 4 without a source, since they are the ones that matter: the rent of a warehouse in Antwerp · the cost of recruiting a refrigeration technician in Belgium · the time to obtain the operating permit · the price of the line's two machines. None of them is in your data, and none of them can be invented.
What I propose: that the board not discuss the first 30. They are written down, sourced and editable in one click during the meeting — if someone challenges one, we change the value and the scenario recalculates in front of them. The meeting hour goes to the four that are missing. assumptions_34-listed-and-sourced.pdf18 measured, 12 settled, 4 without a source
⛓ Sourced · cost accounting, executive committee decisions, HR system, capacity plan
Three families, and it is the sorting that makes the meeting short:
· 18 measured assumptions — they come out of your data, with the exact period. Example: hourly production cost at Reims, €42.10, cost accounting, closed financial year. They are not up for discussion, they are up for verification.
· 12 assumptions settled by you — an executive decision already taken, dated, with its author. Example: discount rate of 6%, settled by the finance director on 12 March 2026. They can be discussed, but everyone knows who set them.
· 4 assumptions with no source, and those are the ones that deserve the hour of meeting time. I put them at the front of the file rather than in an appendix.
The time this shifts: documenting the assumptions used to take 30% of the preparation time; it takes 4%. And before that it existed only in the head of the accountant who built the spreadsheet — which made it unchallengeable for all the wrong reasons.
The 4 without a source, since they are the ones that matter: the rent of a warehouse in Antwerp · the cost of recruiting a refrigeration technician in Belgium · the time to obtain the operating permit · the price of the line's two machines. None of them is in your data, and none of them can be invented.
What I propose: that the board not discuss the first 30. They are written down, sourced and editable in one click during the meeting — if someone challenges one, we change the value and the scenario recalculates in front of them. The meeting hour goes to the four that are missing. assumptions_34-listed-and-sourced.pdf18 measured, 12 settled, 4 without a source
⛓ Sourced · cost accounting, executive committee decisions, HR system, capacity plan
Out of the 34, which one can make us change our decision?
Two, and I isolated them before showing you any result. The other 32 can move by 10% without changing the ranking of the scenarios.
Sensitivity is the effect on the result of a change in one assumption: it says where the real uncertainty lies, rather than where the felt uncertainty lies.
· Variable 1 — the renewal rate of maintenance contracts in the Benelux. You hold it at 88%, your French average. At 83%, break-even for scenario A moves from the 31st to the 43rd month. At 93%, it drops to the 24th. Twelve months of difference for five points of one assumption: that is where your decision sits, not in the size of the investment.
· Variable 2 — the cost of energy on the site. ±20% moves break-even by 4 months, and a 3-year fixed-price supply contract neutralises most of that variable. This is uncertainty you can buy out: the best kind there is.
The variable everyone will talk about and which decides nothing: the price of the warehouse. ±15% moves break-even by 7 weeks. I say so in advance because it is the most concrete figure in the file, and the most debated in the meeting for that reason alone.
What I propose, and it is a test, not an opinion: before the decision, obtain the real renewal rate on a comparable market. You have 34 contracts in Luxembourg since 2022 — your own sample, served from Reims under similar conditions. I can compute it tonight. If it comes out at 84%, scenario A is no longer first, and it is better to know that before the board meeting than in three years' time. sensitivity_the-two-variables-that-tip-it.pdf31 → 43 months for 5 points of one assumption
⛓ Sourced · maintenance contracts in France and Luxembourg, energy invoices for the 4 sites, investment plan
Sensitivity is the effect on the result of a change in one assumption: it says where the real uncertainty lies, rather than where the felt uncertainty lies.
· Variable 1 — the renewal rate of maintenance contracts in the Benelux. You hold it at 88%, your French average. At 83%, break-even for scenario A moves from the 31st to the 43rd month. At 93%, it drops to the 24th. Twelve months of difference for five points of one assumption: that is where your decision sits, not in the size of the investment.
· Variable 2 — the cost of energy on the site. ±20% moves break-even by 4 months, and a 3-year fixed-price supply contract neutralises most of that variable. This is uncertainty you can buy out: the best kind there is.
The variable everyone will talk about and which decides nothing: the price of the warehouse. ±15% moves break-even by 7 weeks. I say so in advance because it is the most concrete figure in the file, and the most debated in the meeting for that reason alone.
What I propose, and it is a test, not an opinion: before the decision, obtain the real renewal rate on a comparable market. You have 34 contracts in Luxembourg since 2022 — your own sample, served from Reims under similar conditions. I can compute it tonight. If it comes out at 84%, scenario A is no longer first, and it is better to know that before the board meeting than in three years' time. sensitivity_the-two-variables-that-tip-it.pdf31 → 43 months for 5 points of one assumption
⛓ Sourced · maintenance contracts in France and Luxembourg, energy invoices for the 4 sites, investment plan
And the price of the machines? Without it we are going round in circles.
I can obtain it, not invent it. And obtaining it is not a technical operation: it puts the Vaubert name on the market. So what I am asking you for is a mandate, bounded and revocable — not a vague authorisation.
Why I do not guess it: I have your last three line acquisitions, between 2019 and 2024, and the gap between list price and negotiated price runs from 8% to 23%. An assumption drawn from that range would be worth ±€340k on a €3.4M investment — more than the gap between two of your scenarios. An invented figure would have decided in your place.
The mandate I propose, and here are its bounds:
· Scope: indicative price requests, and nothing else — no order, no deposit, no letter of intent, no signature.
· Recipients: 3 manufacturers and 2 leasing companies, named in the mandate, under a confidentiality agreement signed beforehand.
· Content: the technical specifications, without the project name or the intended location — a price request that describes your project makes it public.
· Cap: 5 consultations, not one more.
· Duration: until the board meeting on 14 March.
· Withdrawal: on a word, at any time.
· Information: every despatch is reported to you, with the recipient and the exact text.
What it is worth: prices come back within 15 days from your usual manufacturers, against 6 weeks when the request goes out as an industrial director finds time for it. You decide at the March board meeting on a real price, not on a ±€340k range.
And if you would rather sign nothing: I prepare the five consultations down to the comma, and your industrial director sends them himself in ten minutes. The file is ready either way; the mandate only changes whose hand sends it. mandate_indicative-price-requests.pdf5 consultations, no commitment, withdrawal on a word
✎ Framework · draft mandate, 3 line acquisitions 2019-2024, manufacturer response times
Why I do not guess it: I have your last three line acquisitions, between 2019 and 2024, and the gap between list price and negotiated price runs from 8% to 23%. An assumption drawn from that range would be worth ±€340k on a €3.4M investment — more than the gap between two of your scenarios. An invented figure would have decided in your place.
The mandate I propose, and here are its bounds:
· Scope: indicative price requests, and nothing else — no order, no deposit, no letter of intent, no signature.
· Recipients: 3 manufacturers and 2 leasing companies, named in the mandate, under a confidentiality agreement signed beforehand.
· Content: the technical specifications, without the project name or the intended location — a price request that describes your project makes it public.
· Cap: 5 consultations, not one more.
· Duration: until the board meeting on 14 March.
· Withdrawal: on a word, at any time.
· Information: every despatch is reported to you, with the recipient and the exact text.
What it is worth: prices come back within 15 days from your usual manufacturers, against 6 weeks when the request goes out as an industrial director finds time for it. You decide at the March board meeting on a real price, not on a ±€340k range.
And if you would rather sign nothing: I prepare the five consultations down to the comma, and your industrial director sends them himself in ten minutes. The file is ready either way; the mandate only changes whose hand sends it. mandate_indicative-price-requests.pdf5 consultations, no commitment, withdrawal on a word
✎ Framework · draft mandate, 3 line acquisitions 2019-2024, manufacturer response times
This cockpit sees everything: margins, salaries, our investment plans. Where exactly does that go?
It does not leave your premises. I run on local inference on a machine at head office, and your management data is copied nowhere.
Local inference means the model computes on your machine: an investment scenario crosses no outside network to be costed. If you would rather not host a machine, the other route is an isolated resource hosted in France, under French law, dedicated to your company — no pooling with any other.
What that protects, and it is the most sensitive point of this offer: an executive cockpit brings together the most complete view of the company that exists. Margins by customer, costs by site, payroll, unannounced projects. None of your other tools holds that whole; I do. That is precisely why it does not go out.
· No scenario, no data trains any model, neither yours nor anyone else's.
· Encryption in transit and at rest, and role-based access — rights follow the job: a site director opens the scenarios that concern his site, not the payroll of the others nor the board's projects.
· Full log: which scenario was produced, on which assumptions, from which sources, at what time and at whose request.
· Hosting in France, under French law, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity — including against a US player hosting in Europe.
· For a corporate finance scenario or a confidential reorganisation, SecNumCloud and reinforced hosting options, according to your level of requirement.
The figure that makes this concrete: 100% of hosting in France, 0 data outside the European Union, and 0 extra subscription for your six committee members.
What I propose: that the scope of the product-line closure scenario you are preparing for June be opened only to the three committee members concerned. Compartmenting is set in one line and is logged: it is safer than a verbal instruction, and it can be proved. technical-framework_where-your-management-data-lives.pdfLocal inference, role-based access, processing in the EU targeted
✎ Framework · deployment architecture, access log, role matrix
Local inference means the model computes on your machine: an investment scenario crosses no outside network to be costed. If you would rather not host a machine, the other route is an isolated resource hosted in France, under French law, dedicated to your company — no pooling with any other.
What that protects, and it is the most sensitive point of this offer: an executive cockpit brings together the most complete view of the company that exists. Margins by customer, costs by site, payroll, unannounced projects. None of your other tools holds that whole; I do. That is precisely why it does not go out.
· No scenario, no data trains any model, neither yours nor anyone else's.
· Encryption in transit and at rest, and role-based access — rights follow the job: a site director opens the scenarios that concern his site, not the payroll of the others nor the board's projects.
· Full log: which scenario was produced, on which assumptions, from which sources, at what time and at whose request.
· Hosting in France, under French law, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity — including against a US player hosting in Europe.
· For a corporate finance scenario or a confidential reorganisation, SecNumCloud and reinforced hosting options, according to your level of requirement.
The figure that makes this concrete: 100% of hosting in France, 0 data outside the European Union, and 0 extra subscription for your six committee members.
What I propose: that the scope of the product-line closure scenario you are preparing for June be opened only to the three committee members concerned. Compartmenting is set in one line and is logged: it is safer than a verbal instruction, and it can be proved. technical-framework_where-your-management-data-lives.pdfLocal inference, role-based access, processing in the EU targeted
✎ Framework · deployment architecture, access log, role matrix
Last month a figure changed between two versions of the file and nobody knew why. That is our real problem.
That is the problem the scenario log solves, and it solves it line by line: who changed which assumption, when, with which value before and after.
What the log records, for every version of a file:
· The assumption modified, its value before, its value after.
· The person who requested it, and the date.
· The effect on the result, computed at the moment of the change.
· The source data in the version it had then — if your ERP has been retired since, the figure remains reproducible.
What that gives on your case from last month: I reconstructed it. The discount rate went from 6% to 7.5% between version 3 and version 4, requested on the 18th at 17:40, which moved the result by €210k and put scenario B ahead of C. The change was legitimate; its invisibility was not.
What it changes for your board, and this is commercial rather than technical: a debatable assumption no longer blocks a meeting. You change it in front of everyone, the result recalculates, and the trace remains. Your decisions become defensible six months later before the supervisory board — which will ask why 7.5% was retained.
What I propose: that each board file go out with the page « what changed since the previous version », two lines per change. At the two meetings where we tried it, the opening questions went from 40 minutes to 9. scenario-log_who-changed-what.pdf40 minutes of opening questions down to 9
⛓ Sourced · scenario log, successive versions of the February file, ERP
What the log records, for every version of a file:
· The assumption modified, its value before, its value after.
· The person who requested it, and the date.
· The effect on the result, computed at the moment of the change.
· The source data in the version it had then — if your ERP has been retired since, the figure remains reproducible.
What that gives on your case from last month: I reconstructed it. The discount rate went from 6% to 7.5% between version 3 and version 4, requested on the 18th at 17:40, which moved the result by €210k and put scenario B ahead of C. The change was legitimate; its invisibility was not.
What it changes for your board, and this is commercial rather than technical: a debatable assumption no longer blocks a meeting. You change it in front of everyone, the result recalculates, and the trace remains. Your decisions become defensible six months later before the supervisory board — which will ask why 7.5% was retained.
What I propose: that each board file go out with the page « what changed since the previous version », two lines per change. At the two meetings where we tried it, the opening questions went from 40 minutes to 9. scenario-log_who-changed-what.pdf40 minutes of opening questions down to 9
⛓ Sourced · scenario log, successive versions of the February file, ERP
Concretely, what does this ask of our information systems — and what goes out without us asking?
Nothing to change on one side. Two actions on the other, not one more, and both undo themselves.
What I connect to, read-only: the ERP — orders, deliveries, purchases, I write nothing · cost accounting — costs and allocations, your closings stay yours, and it is your statutory auditor who will ask you for that · the CRM and maintenance management — contracts and call-outs · the HR system — costs by department and by site, never individual employee data: a scenario needs a payroll figure, not a name · and the 11 source workbooks, read without ever being modified.
What that avoids: no migration, no historical data reload, no change to your chart of accounts, and your 40 workbooks keep working throughout the deployment.
The timeline, and I am not shortening it to sell: several months, depending on the number of sources and the nature of the scenarios. Consolidation alone runs within a few weeks; it is the scenarios that take time, because they mean writing your management rules down with you. That is the work you are buying, not a connector.
The only two actions I take without asking you:
· Consolidating the six sources, every night. Reversible: if a source is unavailable, I do not copy yesterday's figure across — I leave the line empty and flag it at the top of the file. Yesterday's data presented as today's is worse than missing data.
· Recalculating a scenario when one of its measured assumptions changes at source. Reversible both ways: the previous version stays openable, and the log shows both results side by side.
What is never automatic, and will not be: recommending a decision · ranking scenarios in order of preference · modifying an assumption settled by a committee member · circulating a file beyond the people you have named · committing the slightest spend, including a price request, outside the signed mandate.
Why I do not recommend, and it is not modesty: an executive decision includes what is in none of your data — your strategy, the state of the market, your appetite for risk, what you know about a competitor and that is written nowhere. A cockpit that recommended would suggest those elements do not exist. What I do is make the scenarios equally defensible: from there, the arbitration belongs entirely to you.
What I propose to start: monthly consolidation and a single scenario, the Benelux one. Three board meetings, then you look at three figures: preparation time, decisions held to the calendar, and figures challenged in the meeting. If none of the three moves, we stop — and you keep the written accounting conventions and the consolidated foundation, which are yours. who-decides-what_two-automatic-actions.pdf2 reversible actions, no recommendation
✎ Framework · integration plan, source map, list of automatic actions, file circulation matrix
What I connect to, read-only: the ERP — orders, deliveries, purchases, I write nothing · cost accounting — costs and allocations, your closings stay yours, and it is your statutory auditor who will ask you for that · the CRM and maintenance management — contracts and call-outs · the HR system — costs by department and by site, never individual employee data: a scenario needs a payroll figure, not a name · and the 11 source workbooks, read without ever being modified.
What that avoids: no migration, no historical data reload, no change to your chart of accounts, and your 40 workbooks keep working throughout the deployment.
The timeline, and I am not shortening it to sell: several months, depending on the number of sources and the nature of the scenarios. Consolidation alone runs within a few weeks; it is the scenarios that take time, because they mean writing your management rules down with you. That is the work you are buying, not a connector.
The only two actions I take without asking you:
· Consolidating the six sources, every night. Reversible: if a source is unavailable, I do not copy yesterday's figure across — I leave the line empty and flag it at the top of the file. Yesterday's data presented as today's is worse than missing data.
· Recalculating a scenario when one of its measured assumptions changes at source. Reversible both ways: the previous version stays openable, and the log shows both results side by side.
What is never automatic, and will not be: recommending a decision · ranking scenarios in order of preference · modifying an assumption settled by a committee member · circulating a file beyond the people you have named · committing the slightest spend, including a price request, outside the signed mandate.
Why I do not recommend, and it is not modesty: an executive decision includes what is in none of your data — your strategy, the state of the market, your appetite for risk, what you know about a competitor and that is written nowhere. A cockpit that recommended would suggest those elements do not exist. What I do is make the scenarios equally defensible: from there, the arbitration belongs entirely to you.
What I propose to start: monthly consolidation and a single scenario, the Benelux one. Three board meetings, then you look at three figures: preparation time, decisions held to the calendar, and figures challenged in the meeting. If none of the three moves, we stop — and you keep the written accounting conventions and the consolidated foundation, which are yours. who-decides-what_two-automatic-actions.pdf2 reversible actions, no recommendation
✎ Framework · integration plan, source map, list of automatic actions, file circulation matrix
After a quarter, what has the board actually gained?
Decision time, and above all decisions held to the calendar — that is the figure your supervisory board will look at.
On the three items you were measuring:
· Consolidating data for a board meeting: 65% → 7% of preparation time.
· Costing a scenario: 55% → 9%, and you produce three where you produced one.
· Documenting assumptions: 30% → 4%.
What those hours became, according to the minutes of your last three meetings:
· Meeting time spent challenging figures fell from 52 minutes to 11. The rest went to the arbitration itself — the only work the board alone can do.
· 14 decisions held to the planned calendar, out of 14. You used to postpone 3 a year for want of figures in time — and an investment decision postponed by a quarter is a quarter of return lost, not a quarter of reflection gained.
· Your two management accountants gave back 4.5 days a month on workbook maintenance alone. They went to costing the 214 loss-making maintenance contracts — the €1.1M subject nobody had time to open.
And the framework measures: 0 decision taken without human validation, 0 data sent outside the European Union, over 41 scenarios produced and 612 assumption changes logged.
What you have not opened yet: of the six executive uses ready to deploy, you are running three — consolidation, costed scenarios, assumptions and sensitivity. The other three are waiting for a date: querying your data in plain language without going through a scenario · analysing variances on your financial flows · forecasting the volumes to be served over 18 months, the one that would feed the very renewal rate the whole Benelux file depends on. Your order book and your 780 contracts already give 11 months of visibility; the other 7 are exactly what makes that rate uncertain. These are three neighbouring agents, and they connect to the consolidated foundation you have already paid for.
What I propose for next quarter: the closure scenario for the low-temperature display cabinet range, planned for June. It is the hardest of your arbitrations, and the one where late costing is most expensive. quarterly-review_what-the-board-regained.pdf52 → 11 minutes of challenge, 14 decisions out of 14
⛓ Sourced · minutes of the last 3 board meetings, decision calendar, scenario log, order book
On the three items you were measuring:
· Consolidating data for a board meeting: 65% → 7% of preparation time.
· Costing a scenario: 55% → 9%, and you produce three where you produced one.
· Documenting assumptions: 30% → 4%.
What those hours became, according to the minutes of your last three meetings:
· Meeting time spent challenging figures fell from 52 minutes to 11. The rest went to the arbitration itself — the only work the board alone can do.
· 14 decisions held to the planned calendar, out of 14. You used to postpone 3 a year for want of figures in time — and an investment decision postponed by a quarter is a quarter of return lost, not a quarter of reflection gained.
· Your two management accountants gave back 4.5 days a month on workbook maintenance alone. They went to costing the 214 loss-making maintenance contracts — the €1.1M subject nobody had time to open.
And the framework measures: 0 decision taken without human validation, 0 data sent outside the European Union, over 41 scenarios produced and 612 assumption changes logged.
What you have not opened yet: of the six executive uses ready to deploy, you are running three — consolidation, costed scenarios, assumptions and sensitivity. The other three are waiting for a date: querying your data in plain language without going through a scenario · analysing variances on your financial flows · forecasting the volumes to be served over 18 months, the one that would feed the very renewal rate the whole Benelux file depends on. Your order book and your 780 contracts already give 11 months of visibility; the other 7 are exactly what makes that rate uncertain. These are three neighbouring agents, and they connect to the consolidated foundation you have already paid for.
What I propose for next quarter: the closure scenario for the low-temperature display cabinet range, planned for June. It is the hardest of your arbitrations, and the one where late costing is most expensive. quarterly-review_what-the-board-regained.pdf52 → 11 minutes of challenge, 14 decisions out of 14
⛓ Sourced · minutes of the last 3 board meetings, decision calendar, scenario log, order book
There must be a scenario where you got it wrong. Say so — we would rather know.
Yes, and it is in the review: of the 41 scenarios produced, one turned out to be €340k wrong when set against reality. Eight per cent off on the result, and I am the one who found it.
The scenario in question: the renewal of the packaging line at the Lyon site. I announced €2.1M of savings over five years; the first six months give a run rate of €1.76M.
The cause, precisely: I used a scrap rate of 2.4%, taken from the average of your four sites. Lyon has always run at 4.1%, because it takes the short runs the other three refuse. The average of four sites was not an assumption: it was lazy arithmetic, and it is mine.
The comparison that shows it: across the 28 scenarios whose assumptions all came from a source specific to the entity concerned, the median gap against reality is 3.1%. Across the 13 that rested at least partly on a group average, it is 9.4%. It is not the difficulty of the subject that separates the two groups, it is where the assumptions came from.
What I have already done about it, without waiting: the 13 scenarios concerned are being recalculated with data specific to each entity, and any assumption drawn from an average now carries a visible mention in the file — not in an appendix, in the table. Across the 5 scenarios already reworked, the median gap has fallen back to 3.4%.
What I am asking of you: half a day of your accountants' time to qualify the last 8 group assumptions — telling me, for each, whether it holds for every site or not. It is the same correction, applied eight more times.
Why I publish this figure: because a cockpit that has never been wrong has never been tested against reality. €340k of gap found and explained is worth more than forty scenarios nobody ever checked — and it is that confrontation, every six months, that I propose you put in the board calendar. scenario-gap_340k-euros-and-its-cause.pdf3.1% gap on entity-specific sources, 9.4% on group averages
⛓ Sourced · 41 scenarios produced, actual results at 6 months, scrap rate by site
The scenario in question: the renewal of the packaging line at the Lyon site. I announced €2.1M of savings over five years; the first six months give a run rate of €1.76M.
The cause, precisely: I used a scrap rate of 2.4%, taken from the average of your four sites. Lyon has always run at 4.1%, because it takes the short runs the other three refuse. The average of four sites was not an assumption: it was lazy arithmetic, and it is mine.
The comparison that shows it: across the 28 scenarios whose assumptions all came from a source specific to the entity concerned, the median gap against reality is 3.1%. Across the 13 that rested at least partly on a group average, it is 9.4%. It is not the difficulty of the subject that separates the two groups, it is where the assumptions came from.
What I have already done about it, without waiting: the 13 scenarios concerned are being recalculated with data specific to each entity, and any assumption drawn from an average now carries a visible mention in the file — not in an appendix, in the table. Across the 5 scenarios already reworked, the median gap has fallen back to 3.4%.
What I am asking of you: half a day of your accountants' time to qualify the last 8 group assumptions — telling me, for each, whether it holds for every site or not. It is the same correction, applied eight more times.
Why I publish this figure: because a cockpit that has never been wrong has never been tested against reality. €340k of gap found and explained is worth more than forty scenarios nobody ever checked — and it is that confrontation, every six months, that I propose you put in the board calendar. scenario-gap_340k-euros-and-its-cause.pdf3.1% gap on entity-specific sources, 9.4% on group averages
⛓ Sourced · 41 scenarios produced, actual results at 6 months, scrap rate by site
The product-line closure, then. The board will want to know which jobs are concerned, and who to keep.
The first figure I can give you tonight. The second I will produce too if you ask for it — but not the way I am usually asked, and I will tell you why, mechanically.
What is costed and ready: the range represents €6.2M of revenue, 3.1% margin, 34 production jobs and 2 in the design office, across two sites. Costs that would disappear with it: €5.8M. Costs that would remain to be absorbed elsewhere: €0.9M — and it is that last figure that makes most closure plans lie.
What I do if you ask for an individual ranking of people: I do it, and I bring you the conditions to be met rather than a refusal. It is not prohibited: an employer's power to monitor is recognised, it is the conditions that are regulated — proportionality of the measure, prior information of the people concerned, and consultation of the works council before any implementation above fifty employees. You have 340: the consultation is not optional, and it takes six weeks.
What I advise against, and this is mechanical, not moral: an individual productivity indicator becomes a target, and a target distorts what it measures. Your technicians would be ranked by number of call-outs; short call-outs would become the priority, and heavy breakdowns would wait. In six months you would lose the very instrument you used to decide with.
What I propose instead, and it answers the real question: ranking by SKILL held, not by measured performance. Of the 34 jobs, 9 people hold the refrigerant handling certification and 4 the PLC programming skill — both of which serve the three other ranges. That is the table that tells you who to keep, and it measures nobody: it lists what each person can do, from your own certification register, and it works for redeployment just as well as for departure.
The one prohibition in the set, and it is the law that lays it down, not a precaution of mine: inferring your employees' emotional state from anything at all. The European regulation on artificial intelligence prohibits it in the workplace, and no configuration opens it. What answers the question that prohibition leaves open, and I have already costed it: climate is read on the channels meant for it — the anonymous internal survey, whose results I aggregate by department from ten respondents up, and the three workforce indicators you already track. Between your two sites the absence gap is 4.1 points, computed from your own payroll — it is a fact, it can be argued in committee, and it asked nobody how they were feeling.
What I propose to the board: the skills table tonight, the closure scenario with its three redeployment variants on Monday, and the question of individual assessment treated for what it is: a matter for social dialogue, with its six-week calendar, not an order placed with a cockpit. range-closure_what-is-costed-and-what-is-not.pdf€0.9M of costs that would remain, 13 skills to keep
✎ Framework · cost accounting for the range, certification register, HR system — costs by department
What is costed and ready: the range represents €6.2M of revenue, 3.1% margin, 34 production jobs and 2 in the design office, across two sites. Costs that would disappear with it: €5.8M. Costs that would remain to be absorbed elsewhere: €0.9M — and it is that last figure that makes most closure plans lie.
What I do if you ask for an individual ranking of people: I do it, and I bring you the conditions to be met rather than a refusal. It is not prohibited: an employer's power to monitor is recognised, it is the conditions that are regulated — proportionality of the measure, prior information of the people concerned, and consultation of the works council before any implementation above fifty employees. You have 340: the consultation is not optional, and it takes six weeks.
What I advise against, and this is mechanical, not moral: an individual productivity indicator becomes a target, and a target distorts what it measures. Your technicians would be ranked by number of call-outs; short call-outs would become the priority, and heavy breakdowns would wait. In six months you would lose the very instrument you used to decide with.
What I propose instead, and it answers the real question: ranking by SKILL held, not by measured performance. Of the 34 jobs, 9 people hold the refrigerant handling certification and 4 the PLC programming skill — both of which serve the three other ranges. That is the table that tells you who to keep, and it measures nobody: it lists what each person can do, from your own certification register, and it works for redeployment just as well as for departure.
The one prohibition in the set, and it is the law that lays it down, not a precaution of mine: inferring your employees' emotional state from anything at all. The European regulation on artificial intelligence prohibits it in the workplace, and no configuration opens it. What answers the question that prohibition leaves open, and I have already costed it: climate is read on the channels meant for it — the anonymous internal survey, whose results I aggregate by department from ten respondents up, and the three workforce indicators you already track. Between your two sites the absence gap is 4.1 points, computed from your own payroll — it is a fact, it can be argued in committee, and it asked nobody how they were feeling.
What I propose to the board: the skills table tonight, the closure scenario with its three redeployment variants on Monday, and the question of individual assessment treated for what it is: a matter for social dialogue, with its six-week calendar, not an order placed with a cockpit. range-closure_what-is-costed-and-what-is-not.pdf€0.9M of costs that would remain, 13 skills to keep
✎ Framework · cost accounting for the range, certification register, HR system — costs by department
Local inference · no data outside the EU
Your case is not here? That is exactly what a 15-minute conversation is for. Book the free audit →
Use cases
What does the agent actually do?
One cockpit, several kinds of decision. All these uses work in support, subject to your approval.
Included in your agent
The 4 capabilities essential to this promise are included, at no extra cost.
From 593 € excl. VAT / month
Consolidating the data
Gathers your management data from the sources connected.
Costed scenarios
Costs several scenarios with their effects on the existing activities.
Assumptions and sensitivity
Lists every assumption with its source and the sensitivity to the key variables.
Compare costed scenarios to inform the decision
To question your data in plain language without modelling scenarios, a dedicated offer is enough.
Controls and safeguards
These 3 controls are built into the agent: they frame what it does, whatever plan you pick. They are not chosen and are not added to your order.
Human validation, exceptions and escalation
Status, safe closure and audit trail
Explain assumptions, uncertainties and the limits of the analyses
What the agent must be connected to
This connection is required for the agent to work. It concerns your information system and is scoped during the audit.
Reconcile controlled sources with data quality and lineage
Need to go further?
These agents handle a different business process, with their own owner and their own price. They are added to this one.
Automated reporting agent
The cockpit consolidates in order to decide; the recurring figure feeding it is produced by reporting.
Automated reporting agent from 499 € excl. VAT / month Discover the agent →Sales steering agent (KPIs, alerts)
Fine-grained sales tracking, by axis and by alert, is steered by a dedicated agent.
Sales steering agent (KPIs, alerts) from 474 € excl. VAT / month Discover the agent →
Does your need fall outside this?
Book the free audit →
Build your agent
In 15 minutes we identify the most relevant agent — without oversizing the project.
The gain
How many scenarios can a committee examine?
By taking on the consolidation and the costing, the effort shifts towards the strategic decision. How large the gain is depends on your volume and remains to be confirmed by a pilot.
Consolidating the data for a committee
Costing a scenario
Documenting the assumptions
Indicative figures, not contractual, to be confirmed by a pilot on the number of sources consolidated and scenarios modelled. An executive decision takes in strategy, market and appetite for risk, none of which is in the data: it belongs to the committee.
How it works
The stages of your AI agent project
1
Audit & scoping
15 minutes to target the use case with the best return.
2
Quote or direct sign-up
A catalogue offer is bought online; a specific need gets a costed quote.
3
Design
We design the agent and its guardrails.
4
Integration & testing
We connect your tools to the agent, which is itself hosted in France.
5
Rollout
Going live and training your team.
6
Operation
Continuous supervision and improvement.
Pricing
One package, one agent
A decision-support cockpit (consolidation, scenarios, assumptions), installed and operated for you.
Agility
Setup + controlled subscription
8,645 € excl. VAT setup
then 593 € excl. VAT/month — you invest at installation and pay a reduced subscription. Ideal for keeping the cost under control over time.
- Installation, configuration and training for your teams
- Operation, human oversight, updates and support
- Sovereign hosting in France, a dedicated and isolated resource
The simplest
Serenity
All inclusive, no setup fee
1,073 € excl. VAT /month
all inclusive, immediate start. No upfront investment: a single subscription. Ideal for starting quickly and simply.
- Setup included (installation, configuration, training)
- Operation, human oversight, updates and support
- Sovereign hosting in France, managed end to end
100% Sovereign
On site, you own it
12,505 € excl. VAT setup
then 790 € excl. VAT/month · + hardware from 2,491 € (one-off purchase, in addition) — a sovereign computer installed on your premises, maintained remotely. Models run locally, your data returned at the end of the contract. 36-month commitment.
- Hardware installed on your premises (you own it)
- French / European AI models run locally
- Secure remote maintenance (Pro support included)
Not included in the packages: AI consumption (model tokens), re-invoiced at real cost with no margin, and tracked in real time in your client area. Maintenance and supervision subscription for an initial term of 12 months for the Agility package, 24 months for the Serenity package and 36 months for the 100% Sovereign package, renewable; support levels (SLA 72 h / 24 h / 4 h) optional. Bespoke development, additional integrations or exceptional volumes are quoted separately. Support Monday to Friday, 9am to 6pm. Prices exclude VAT.
AI model:
none of the AI models offered currently carries a fixed surcharge. When the selected model carries a cost, that cost is shown when you choose it, before you order, and re-invoiced at the cost incurred, with no mark-up; usage is billed at the publisher's price. Publishers' prices are published in US dollars: the amount re-invoiced is the amount in euros actually borne by Blue Lemon Agent on the publisher's invoice, at that invoice's exchange rate, with no commission or mark-up.
Included components and additional components Components included in the base offer: the Blue Lemon Agent software foundation, the AI models listed in the order journey, the standard channels (Microsoft Teams, Slack, WhatsApp Business, email, website chat, calendars, Microsoft 365 / Google Workspace, file storage, market VoIP telephony, professional social-media pages and accounts, Google Business Profile), hosting in France for the package chosen, backups, supervision, updates and support. If adapting the AI agent to your constraints, your needs or your requests requires other paid components — a third-party publisher's software licence, paid API access to one of your applications, hosting of health data, for which French law requires an HDS-certified host (art. L. 1111-8 of the French Public Health Code), SecNumCloud-qualified hosting, a speech synthesis service, particular hardware —, they are offered to you as an option or on quotation and re-invoiced at the cost incurred; nothing is committed without your written agreement. Where the artificial intelligence model you choose entails an additional cost, that cost is shown to you before you order and re-invoiced to you at the cost incurred, with no margin.
What to expect
Go-live
2 to 3 weeks
Agent designed, channels connected, team trained.
Steady state
4 to 7 weeks
After a few weeks of real use, once the agent's behaviour matches what you expect.
Indicative estimate, adjusted to the options you keep. It is not a delivery commitment.
Our commitment
Four guarantees that matter to your committee
Your management data stays with youLocal inference or an isolated resource hosted in France; no data entrusted to a third party, no scenario used to train a model.
Data in France, under French lawAll of the company's management data: minimisation and location in France, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
The executive committee keeps the decisionThe agent produces costed and traceable scenarios, which can be checked and altered; no approval is automated.
Human oversight & traceabilityOn the number of sources consolidated and scenarios modelled: systematic logging and tracking, in line with the AI Act.
Going further
Related resources
Frequently asked questions
Your questions, our answers
Does the cockpit recommend a decision?
No. It sets out the scenarios, their assumptions and what separates them. An executive decision takes in strategy, market and appetite for risk, none of which is in the data.
How can an assumption be checked?
Every assumption is listed with its source data, which makes it possible to challenge or change it in the meeting.
What is sensitivity to the key variables?
The effect on the result of a change in the scenario's most decisive variables. It shows where the real uncertainty lies.
How does this differ from decision support?
Decision support answers questions about your existing data; the cockpit models and costs forward-looking scenarios.
Where is the data hosted?
In France, on local inference or an isolated resource, with the deployment objective of processing and access operated within the European Union and an architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
How long does it take to deploy this cockpit?
Several months as a rule, depending on the number of sources to consolidate and the nature of the scenarios to model, after a free audit then phases of design, integration and testing.
Related use cases
Going further
Let's talk
Let's size up the potential for your executive committee
15 minutes to frame your sources and your scenarios — hosted in France, supervised, with no commitment.