● B2B offer — Logistics & optimisation
Logistics optimisation: several plans, your constraints met
Organising flows between several sites means making capacities, promised lead times and transport costs all fit. Your agent assembles those constraints across the whole network and proposes several complete plans, each optimising an explicit criterion and stating what it gives up elsewhere. Hosted in France: your logistics costs and your load plans stay with you. The logistics department picks the plan and commits the resources.
Hosted in France
Logistics costs protected
GDPR & AI Act: governed deployment
Human oversight
Updated on
Deployed in a few weeks
Logistics optimisation · hosted in France
Propose an arrangement of flows for next month.
Three complete plans across the whole network, respecting each site's capacity and the promised lead times.
For each: the criterion optimised — transport cost, average lead time, robustness to disruption — and what it gives up elsewhere.
The constraints used are listed with each plan.
🔗 Sourced · capacities, lead times and costs from your systems
For each: the criterion optimised — transport cost, average lead time, robustness to disruption — and what it gives up elsewhere.
The constraints used are listed with each plan.
🔗 Sourced · capacities, lead times and costs from your systems
Does the cheapest plan hold up if something goes wrong?
Its margin is tighter on two links: the points concerned and the size of the margin are stated.
Choosing between cost and robustness is a matter of your service policy: that call is yours, as is committing the resources.
✎ Support · margins set out, logistics decision
Choosing between cost and robustness is a matter of your service policy: that call is yours, as is committing the resources.
✎ Support · margins set out, logistics decision
Local inference · no data outside the EU
Plans 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 logistics department decidesThe agent prepares, never rules
✦ In brief
A Blue Lemon Agent logistics optimisation agent assembles capacities, lead times and costs across your whole network and proposes several complete plans, each optimising an explicit criterion and stating what it gives up elsewhere. Choosing the plan and committing the resources stay with the department. It runs on local inference or is hosted in France: your logistics costs stay with you, 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
logistics 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 your number of sites, links and constraints is confirmed by a pilot.
The context
What does an AI agent bring to your logistics arrangement?
Comparing several complete arrangements rather than adjusting the last one: that is where the logistics gains are.
! The issue
A logistics plan makes capacities by site, promised lead times and transport costs fit together. Exploring several complete arrangements by hand is out of reach; so you adjust the previous one. The agent builds several, each optimising an explicit criterion, with the constraints used and the margins obtained.
✓ Our answer
The logistics department compares complete plans and decides between cost, lead time and robustness according to its service policy. Committing resources — chartering, opening a link, adding capacity — remains its decision. Local inference or an isolated resource hosted in France: your logistics costs and your load plans, information a carrier could use in a negotiation, do not leave the company.
The decisive point
Your logistics costs and your load plans: sovereignty & compliance
Your logistics costs and your load plans weigh in any negotiation with a carrier. Here is how they are protected.
Local inference
The agent can run on a machine belonging to your organisation: no logistics cost and no load plan leaves the network.
Hosting in France
Otherwise, a dedicated and isolated resource hosted in France, under French law — your flows, your sites and your transport constraints: processing and access within the European Union targeted by the architecture.
Reduced extraterritorial exposure
For your logistics costs and your load plans, 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 logistics network and its constraints.
Constraints and margins shown
Every plan states the constraints used, the criterion optimised and the margins obtained; encryption, role-based access and logging of the plans produced.
AI Act: governed deployment
The agent is strictly in support; no resource is committed and no link is opened automatically; 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 flows involving sensitive goods 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 companyArdenor — wholesale distributor of electrical equipment to trade professionals
- Sector
- Distribution of electrical equipment to installers and construction firms
- Headcount
- 1,100 staff, including 9 in logistics: 1 director, 3 planners, 5 operations controllers
- Network served
- 3 distribution centres — North, Centre, South-East —, 22 branches and 41 regular lanes
- Volume handled
- 18,000 order lines a week, 1.4 million movements over 18 months, €6.2M of transport cost a year
- Tools in place
- Warehouse system, transport system, ERP and a load-plan spreadsheet — the agent plugs into them, nothing is replaced
- Who decides
- The logistics director settles the plan and commits the means; centre managers approve capacities; sales own the promised lead times
- Room for improvement
- A complete plan takes 3 weeks and exists in a single version; building it takes 70 % of the exercise, and margins per lane are measured nowhere
Ardenor adjusts last month's plan every month for want of time to build another, and overruns one centre's capacity one week in three. The agent runs on local inference on a machine at the company and reads the warehouse system, the transport system, the ERP and the load-plan spreadsheet: it assembles the constraints, builds several complete plans, exposes the margins and costs what each one concedes. The logistics director settles the plan and signs the mandate. The exchanges below cover six months, from the first plan to the review.
This company, its figures and the exchanges that follow were invented for the demonstration. They illustrate a common situation; they describe no real client.
Logistics optimisation · hosted in France
I read eighteen months of your flows last night — 1.4 million movements, 41 lanes, 3 centres, 22 branches — and I bring you three complete network organisations, not three variants of today's.
A lane is a regular run between two points of your network: a distribution centre and a branch, or two centres.
Three measured gaps, and they carry most of the gain:
· 11 lanes out of 41 carry 74 % of your transport cost. Your logistics is decided on eleven runs, not on forty-one.
· Two lanes run at 51 % fill, every day, for eighteen months. One trailer in two leaves half empty on those two runs.
· 63 tonnes a year of identical part numbers travel in both directions, North to South and South to North, within the same week. Nobody could see it: each movement is justified on its own, and nothing brings them together.
The three plans, on the same constraints and the same volumes:
· Cost plan — €5.74M a year. Concedes: 6 lanes below 4 % of capacity margin, and an average lead time of 1.9 days.
· Lead-time plan — €6.08M. Average lead time 1.4 days, commitments met 99.1 %. Concedes: €340,000 more.
· Robustness plan — €5.91M. No critical lane below 9 % margin, two carriers on every major lane. Concedes: €170,000 more than the cost plan.
The time this moves: building a complete plan took 70 % of the exercise; it takes 9 %. Three weeks of work become one day, and you compare three organisations where you used to adjust the previous one.
What I propose: before you choose, I rebuild the three plans on the real capacities of your centres — and one of the three is not what your spreadsheet declares. logistics-plans_three-organisations.pdf€5.74 / 6.08 / 5.91M, costed concessions
⛓ Sourced · warehouse system (18 months, 1.4 million movements), transport system, ERP, load-plan spreadsheet
A lane is a regular run between two points of your network: a distribution centre and a branch, or two centres.
Three measured gaps, and they carry most of the gain:
· 11 lanes out of 41 carry 74 % of your transport cost. Your logistics is decided on eleven runs, not on forty-one.
· Two lanes run at 51 % fill, every day, for eighteen months. One trailer in two leaves half empty on those two runs.
· 63 tonnes a year of identical part numbers travel in both directions, North to South and South to North, within the same week. Nobody could see it: each movement is justified on its own, and nothing brings them together.
The three plans, on the same constraints and the same volumes:
· Cost plan — €5.74M a year. Concedes: 6 lanes below 4 % of capacity margin, and an average lead time of 1.9 days.
· Lead-time plan — €6.08M. Average lead time 1.4 days, commitments met 99.1 %. Concedes: €340,000 more.
· Robustness plan — €5.91M. No critical lane below 9 % margin, two carriers on every major lane. Concedes: €170,000 more than the cost plan.
The time this moves: building a complete plan took 70 % of the exercise; it takes 9 %. Three weeks of work become one day, and you compare three organisations where you used to adjust the previous one.
What I propose: before you choose, I rebuild the three plans on the real capacities of your centres — and one of the three is not what your spreadsheet declares. logistics-plans_three-organisations.pdf€5.74 / 6.08 / 5.91M, costed concessions
⛓ Sourced · warehouse system (18 months, 1.4 million movements), transport system, ERP, load-plan spreadsheet
Where do you get our centre capacities from? We overrun them one week in three.
I measured them rather than copying them, and that is exactly why the plan breaks one week in three: the Centre site is declared at 3,200 parcels a day and has held that rate on only 4 days out of 390.
Sustainable capacity is the volume a site holds day after day without degrading its lead times or resorting to overtime — not its record.
What eighteen months of production say, site by site:
· Centre — declared 3,200 parcels/day, sustainable 2,750. Above 2,750, picking time moves from 4 h to 9 h and the picking error rate doubles. All 4 days at 3,200 were followed by a catch-up the next day.
· North — declared 4,000, sustainable 4,150. It is worth more than the spreadsheet credits it with, and that margin of 150 parcels is used in all three plans.
· South-East — declared 2,400, sustainable 2,380. Almost right.
What the correction changes: I rebuilt the three plans on those real capacities. The cost plan moves from €5.74M to €5.79M — €50,000 more, and a plan that holds. Those €50,000 were already being paid, in overtime and catch-ups: they simply never appeared in the plan.
The time this moves: checking site capacities took 40 % of the exercise; it takes 5 %, and it covers 390 real days instead of a figure declared once and for all.
What I propose: that sustainable capacities be recomputed every quarter and approved by your three centre managers. They are their figures, measured at their sites — the argument about Centre's capacity has been running for two years and it is settled in one meeting. sustainable-capacities_3-centres.pdfCentre: 3,200 declared, 2,750 sustainable
⛓ Sourced · warehouse system (390 production days per centre), picking times, overtime, error rates
Sustainable capacity is the volume a site holds day after day without degrading its lead times or resorting to overtime — not its record.
What eighteen months of production say, site by site:
· Centre — declared 3,200 parcels/day, sustainable 2,750. Above 2,750, picking time moves from 4 h to 9 h and the picking error rate doubles. All 4 days at 3,200 were followed by a catch-up the next day.
· North — declared 4,000, sustainable 4,150. It is worth more than the spreadsheet credits it with, and that margin of 150 parcels is used in all three plans.
· South-East — declared 2,400, sustainable 2,380. Almost right.
What the correction changes: I rebuilt the three plans on those real capacities. The cost plan moves from €5.74M to €5.79M — €50,000 more, and a plan that holds. Those €50,000 were already being paid, in overtime and catch-ups: they simply never appeared in the plan.
The time this moves: checking site capacities took 40 % of the exercise; it takes 5 %, and it covers 390 real days instead of a figure declared once and for all.
What I propose: that sustainable capacities be recomputed every quarter and approved by your three centre managers. They are their figures, measured at their sites — the argument about Centre's capacity has been running for two years and it is settled in one meeting. sustainable-capacities_3-centres.pdfCentre: 3,200 declared, 2,750 sustainable
⛓ Sourced · warehouse system (390 production days per centre), picking times, overtime, error rates
Does the cheapest plan hold if a carrier lets us down?
I show you exactly where it is thin, and I have already built its tightened version: €38,000 a year to remove the three dependencies that cost you 14 disruptions in eighteen months.
Capacity margin is the gap between the volume planned on a lane and what that lane can absorb before it overflows.
The cost plan's margins, lane by lane:
· 6 lanes below 4 % margin, of which 3 below 2 %. On those three, one extra pallet puts you outside the plan.
· 3 critical lanes depend on a single carrier, and those three lanes carry 31 % of the network's volume.
· The other 35 have more than 9 % margin and call for no action. The subject fits in six lines, not forty-one.
What your history says about the risk: 14 disruptions in eighteen months — 6 breakdowns, 5 carrier failures, 3 dock closures. Recorded cost: €96,000 in emergency haulage and missed deliveries, of which €71,000 on the three single-carrier lanes.
The tightened cost plan, already built:
· A second carrier on the 3 critical lanes, volume split 70/30.
· Margin raised to 8 % minimum on the 6 thin lanes.
· Cost: €5.83M, that is + €38,000 a year on the corrected cost plan.
· Effect measured on the 14 replayed disruptions: €71,000 of extra cost would have become €19,000. €38,000 of spend against €52,000 of risk, on your own disruptions.
The time this moves: assessing margins took 30 % of the exercise; it takes 6 %, and it covers all 41 lanes instead of the three there used to be time for.
What I propose: you arbitrate between the four plans in committee, and I first show you what each one does to your promised lead times. margins-per-lane_41-lanes.pdf6 lanes below 4 %, €38,000 against €52,000 of risk
⛓ Sourced · transport system (18 months of spot haulage and disruptions), lane capacities, carrier contracts
Capacity margin is the gap between the volume planned on a lane and what that lane can absorb before it overflows.
The cost plan's margins, lane by lane:
· 6 lanes below 4 % margin, of which 3 below 2 %. On those three, one extra pallet puts you outside the plan.
· 3 critical lanes depend on a single carrier, and those three lanes carry 31 % of the network's volume.
· The other 35 have more than 9 % margin and call for no action. The subject fits in six lines, not forty-one.
What your history says about the risk: 14 disruptions in eighteen months — 6 breakdowns, 5 carrier failures, 3 dock closures. Recorded cost: €96,000 in emergency haulage and missed deliveries, of which €71,000 on the three single-carrier lanes.
The tightened cost plan, already built:
· A second carrier on the 3 critical lanes, volume split 70/30.
· Margin raised to 8 % minimum on the 6 thin lanes.
· Cost: €5.83M, that is + €38,000 a year on the corrected cost plan.
· Effect measured on the 14 replayed disruptions: €71,000 of extra cost would have become €19,000. €38,000 of spend against €52,000 of risk, on your own disruptions.
The time this moves: assessing margins took 30 % of the exercise; it takes 6 %, and it covers all 41 lanes instead of the three there used to be time for.
What I propose: you arbitrate between the four plans in committee, and I first show you what each one does to your promised lead times. margins-per-lane_41-lanes.pdf6 lanes below 4 %, €38,000 against €52,000 of risk
⛓ Sourced · transport system (18 months of spot haulage and disruptions), lane capacities, carrier contracts
Our promised lead times come before cost. That is not negotiable.
The four commitments are hard constraints in all three plans, and I measured how well you actually hold them: 96.2 % overall, but 84.1 % on three branches — and the cause is one hour on a clock.
Your four commitments, as written in your commercial terms: next-day delivery on the core catalogue for the 22 branches · day-after on the extended catalogue · in-branch collection within 4 hours · site delivery in a confirmed slot.
What eighteen months of measurement give:
· Overall performance: 96.2 %, across 1.4 million movements.
· 19 branches out of 22 hold 97 % or better.
· 3 branches at 84.1 %, and they concentrate 61 % of your lead-time complaints.
The cause, and it is mechanical: those three branches are served by the Centre → East lane, which leaves at 19:00. Their commercial cut-off is 18:00 — the latest hour at which an order is taken for next-day delivery. An order placed at 17:50 reaches picking at 18:10 and misses the departure by an hour. It is not a question of means, it is ten minutes between two schedules nobody had ever compared.
The two corrections, both costed:
· Move those three branches' cut-off to 17:30: performance 98.3 %, no cost. Commercial effect measured on history: 2.1 % of the orders concerned would move to the next day.
· Move the lane's departure to 20:00: performance 99.4 %, cost €26,000 a year in shifted picking hours, and no order lost.
My recommendation, and the arithmetic is yours: the second. Those 2.1 % of orders represent €210,000 of revenue displaced at those three branches; €26,000 keeps it same-day. The decision is yours, and it is taken on those two lines. lead-time-performance_22-branches.pdf96.2 % overall, 84.1 % on 3 branches, 2 corrections
⛓ Sourced · commercial terms, warehouse and transport systems (18 months of promised and actual dates), customer complaints
Your four commitments, as written in your commercial terms: next-day delivery on the core catalogue for the 22 branches · day-after on the extended catalogue · in-branch collection within 4 hours · site delivery in a confirmed slot.
What eighteen months of measurement give:
· Overall performance: 96.2 %, across 1.4 million movements.
· 19 branches out of 22 hold 97 % or better.
· 3 branches at 84.1 %, and they concentrate 61 % of your lead-time complaints.
The cause, and it is mechanical: those three branches are served by the Centre → East lane, which leaves at 19:00. Their commercial cut-off is 18:00 — the latest hour at which an order is taken for next-day delivery. An order placed at 17:50 reaches picking at 18:10 and misses the departure by an hour. It is not a question of means, it is ten minutes between two schedules nobody had ever compared.
The two corrections, both costed:
· Move those three branches' cut-off to 17:30: performance 98.3 %, no cost. Commercial effect measured on history: 2.1 % of the orders concerned would move to the next day.
· Move the lane's departure to 20:00: performance 99.4 %, cost €26,000 a year in shifted picking hours, and no order lost.
My recommendation, and the arithmetic is yours: the second. Those 2.1 % of orders represent €210,000 of revenue displaced at those three branches; €26,000 keeps it same-day. The decision is yours, and it is taken on those two lines. lead-time-performance_22-branches.pdf96.2 % overall, 84.1 % on 3 branches, 2 corrections
⛓ Sourced · commercial terms, warehouse and transport systems (18 months of promised and actual dates), customer complaints
There are also driving times, customer receiving windows, the docks. None of that is written down anywhere.
It has been since this morning: 118 constraints formalised, 41 of which lived only in the heads of two of your controllers — and they now carry their names.
What the plan holds together:
· The driving and rest times that road transport regulation imposes on drivers, applied run by run, breaks included.
· Your customers' receiving windows: read off 18 months of refusals and redeliveries, not declared. 34 customers have a narrower window than your transport system assumed, and that explains 210 redeliveries a year at €41 each — €8,600.
· Dock slots per centre: number of docks, measured average loading time, overlaps.
· Product compatibilities, including items subject to particular transport rules, which do not travel with just anything.
· Return runs and equipment collections, which take up 14 % of your kilometres and appeared in no plan.
The 41 constraints written down nowhere: « this customer takes no delivery on Friday after 14:00 », « dock 3 at Centre takes no articulated lorries », « the Vaubourg branch closes between 12:00 and 14:00 ». Your two controllers applied them from memory, and the plan ignored them. Three weeks of absence by one of them last year cost €11,000 in redeliveries.
What formalising changes, measured: replaying the 18 months with all 118 constraints, redeliveries fall from 210 to 48 a year — €6,600 a year, and above all 162 customer appointments kept.
What I propose: every constraint carries the name of whoever supplied it and the date it was last verified. It is their knowledge, written down and attributed — and it no longer goes on holiday with them. network-constraints_118-formalised.pdf41 tacit constraints written down, 210 → 48 redeliveries
⛓ Sourced · transport system (refusals and redeliveries over 18 months), dock plans, operations interviews, applicable transport rules
What the plan holds together:
· The driving and rest times that road transport regulation imposes on drivers, applied run by run, breaks included.
· Your customers' receiving windows: read off 18 months of refusals and redeliveries, not declared. 34 customers have a narrower window than your transport system assumed, and that explains 210 redeliveries a year at €41 each — €8,600.
· Dock slots per centre: number of docks, measured average loading time, overlaps.
· Product compatibilities, including items subject to particular transport rules, which do not travel with just anything.
· Return runs and equipment collections, which take up 14 % of your kilometres and appeared in no plan.
The 41 constraints written down nowhere: « this customer takes no delivery on Friday after 14:00 », « dock 3 at Centre takes no articulated lorries », « the Vaubourg branch closes between 12:00 and 14:00 ». Your two controllers applied them from memory, and the plan ignored them. Three weeks of absence by one of them last year cost €11,000 in redeliveries.
What formalising changes, measured: replaying the 18 months with all 118 constraints, redeliveries fall from 210 to 48 a year — €6,600 a year, and above all 162 customer appointments kept.
What I propose: every constraint carries the name of whoever supplied it and the date it was last verified. It is their knowledge, written down and attributed — and it no longer goes on holiday with them. network-constraints_118-formalised.pdf41 tacit constraints written down, 210 → 48 redeliveries
⛓ Sourced · transport system (refusals and redeliveries over 18 months), dock plans, operations interviews, applicable transport rules
And the day a constraint changes?
It is changed in one word, and the effect on cost, lead time and margins appears in under a minute. I ran three examples you have been debating for months.
Three simulations, ready:
· Open an extra dock at Centre on Saturday mornings. Effect: − €112,000 a year, because the Monday smoothing disappears and Centre's sustainable capacity rises from 2,750 to 3,050 parcels a day midweek. Cost of the dock: €34,000 a year. Net gain: €78,000.
· Move the 3 South-East lanes to a 05:00 departure instead of 06:00. Effect: + 0.4 points of lead-time performance, no cost, capacity margin unchanged. It is the cheapest change of the lot, and it had never been costed.
· Close the Wednesday North → Centre lane, whose average fill is 51 %. Effect: − €61,000 a year, and 2 branches move from next-day to day-after on Thursdays. That one touches your commercial promise: it is an arbitration, not a setting.
What is kept on every change: who changed what, when, the value before and after, and the costed effect observed afterwards. A changed constraint is measured like everything else.
What I propose: open the simulator to your three planners as a self-service tool. They test an organisation in two minutes instead of spending a week on it, and they argue with figures in committee instead of defending intuitions. Every simulation stays a draft until you settle the plan: nothing reaches the transport system without your word.
Three simulations, ready:
· Open an extra dock at Centre on Saturday mornings. Effect: − €112,000 a year, because the Monday smoothing disappears and Centre's sustainable capacity rises from 2,750 to 3,050 parcels a day midweek. Cost of the dock: €34,000 a year. Net gain: €78,000.
· Move the 3 South-East lanes to a 05:00 departure instead of 06:00. Effect: + 0.4 points of lead-time performance, no cost, capacity margin unchanged. It is the cheapest change of the lot, and it had never been costed.
· Close the Wednesday North → Centre lane, whose average fill is 51 %. Effect: − €61,000 a year, and 2 branches move from next-day to day-after on Thursdays. That one touches your commercial promise: it is an arbitration, not a setting.
What is kept on every change: who changed what, when, the value before and after, and the costed effect observed afterwards. A changed constraint is measured like everything else.
What I propose: open the simulator to your three planners as a self-service tool. They test an organisation in two minutes instead of spending a week on it, and they argue with figures in committee instead of defending intuitions. Every simulation stays a draft until you settle the plan: nothing reaches the transport system without your word.
We are going with the tightened robustness plan. What actually happens now?
Eight minutes after your signature. The month's 1,244 transport orders are built in your transport system, checked, and each of the 41 spot movements has two costed quotes against it.
Spot haulage is the one-off use of a carrier for a run the regular plan does not cover.
What is ready:
· 1,244 transport orders, with lane, dock slot, vehicle, intended driver, applicable customer constraints and the promised lead time attached.
· 41 spot movements, each with two quotes from referenced carriers and the price gap. Average gap between best and second: 14 %.
· 3 lane openings, which bind the company for twelve months: they stay for your signature and are presented separately, with their monthly payback.
The time it hands back: your five controllers used to key those orders in 46 hours — more than a week at a full desk. That time goes back into the day's operations, the only place they cannot be replaced.
The mandate I propose, written, capped, dated:
· Scope: the carriers referenced in the framework contract, and nothing else.
· Per-movement cap: €4,500. Above that, it comes to you — 3 of the 41 are in that case.
· Monthly envelope: €180,000, the amount of the plan retained.
· Choice rule: the best of at least two quotes, at identical service and lead time, with the second quote kept on file.
· Term: the current month, renewed in one word.
· Withdrawal: in one word, at any moment, including from your phone.
· Trail: which movement, which carrier, which quotes compared, at which second.
You sign, the 38 covered movements go out in eight minutes, and you keep the 3 large ones and the 3 lane openings in front of you. haulage-mandate_capped-dated.pdf€4,500 per movement, €180,000 envelope
✎ Framework · plan retained, carrier framework contracts, quotes received, mandate settings
Spot haulage is the one-off use of a carrier for a run the regular plan does not cover.
What is ready:
· 1,244 transport orders, with lane, dock slot, vehicle, intended driver, applicable customer constraints and the promised lead time attached.
· 41 spot movements, each with two quotes from referenced carriers and the price gap. Average gap between best and second: 14 %.
· 3 lane openings, which bind the company for twelve months: they stay for your signature and are presented separately, with their monthly payback.
The time it hands back: your five controllers used to key those orders in 46 hours — more than a week at a full desk. That time goes back into the day's operations, the only place they cannot be replaced.
The mandate I propose, written, capped, dated:
· Scope: the carriers referenced in the framework contract, and nothing else.
· Per-movement cap: €4,500. Above that, it comes to you — 3 of the 41 are in that case.
· Monthly envelope: €180,000, the amount of the plan retained.
· Choice rule: the best of at least two quotes, at identical service and lead time, with the second quote kept on file.
· Term: the current month, renewed in one word.
· Withdrawal: in one word, at any moment, including from your phone.
· Trail: which movement, which carrier, which quotes compared, at which second.
You sign, the 38 covered movements go out in eight minutes, and you keep the 3 large ones and the 3 lane openings in front of you. haulage-mandate_capped-dated.pdf€4,500 per movement, €180,000 envelope
✎ Framework · plan retained, carrier framework contracts, quotes received, mandate settings
And the day a lorry breaks down at 4 in the morning?
You have three costed answers at 04:02, and the whole plan is rebuilt in 90 seconds instead of three hours forty.
What I measured across your 14 disruptions of the last eighteen months: 3 h 40 on average to rebuild a day's plan by hand, and in 9 cases out of 14 the solution chosen was not the best available — it was simply the first one found at 5 in the morning.
The typical disruption, replayed: a rigid lorry breaks down on the Centre → South lane at 04:00. 214 parcels on board, 3 branches served, one of them with a site delivery in a confirmed slot.
· Answer A — replacement haulage: available at 05:30, cost €1,340, every lead time met.
· Answer B — transfer onto the same day's North → South lane, which has 17 % margin: cost €0, 2 branches served on time, the third at 14:00 instead of 10:00. The site slot is missed.
· Answer C — split across both: cost €480, every lead time met, site slot included.
My recommendation: C. €480 against €1,340, for the same customer outcome. It could only be found by looking at the margins of all 41 lanes at once, which no controller can do at 4 in the morning.
What I do meanwhile: I have drafted the notice messages to the 3 branches and identified the customer holding the site slot. They go out as soon as you pick an answer.
What that is worth, annualised: across the 14 replayed disruptions, choosing the best answer rather than the first would have saved €34,000 and kept 11 more customer commitments.
What I propose: that I build the three answers automatically as soon as a disruption is logged in the transport system, and that you simply pick one. The choice stays with the controller on duty, and it takes them thirty seconds. replanning_three-costed-answers.pdf90 seconds against 3 h 40, €34,000 a year
⛓ Sourced · transport system (14 disruptions over 18 months), lane margins, haulage quotes, customer commitments
What I measured across your 14 disruptions of the last eighteen months: 3 h 40 on average to rebuild a day's plan by hand, and in 9 cases out of 14 the solution chosen was not the best available — it was simply the first one found at 5 in the morning.
The typical disruption, replayed: a rigid lorry breaks down on the Centre → South lane at 04:00. 214 parcels on board, 3 branches served, one of them with a site delivery in a confirmed slot.
· Answer A — replacement haulage: available at 05:30, cost €1,340, every lead time met.
· Answer B — transfer onto the same day's North → South lane, which has 17 % margin: cost €0, 2 branches served on time, the third at 14:00 instead of 10:00. The site slot is missed.
· Answer C — split across both: cost €480, every lead time met, site slot included.
My recommendation: C. €480 against €1,340, for the same customer outcome. It could only be found by looking at the margins of all 41 lanes at once, which no controller can do at 4 in the morning.
What I do meanwhile: I have drafted the notice messages to the 3 branches and identified the customer holding the site slot. They go out as soon as you pick an answer.
What that is worth, annualised: across the 14 replayed disruptions, choosing the best answer rather than the first would have saved €34,000 and kept 11 more customer commitments.
What I propose: that I build the three answers automatically as soon as a disruption is logged in the transport system, and that you simply pick one. The choice stays with the controller on duty, and it takes them thirty seconds. replanning_three-costed-answers.pdf90 seconds against 3 h 40, €34,000 a year
⛓ Sourced · transport system (14 disruptions over 18 months), lane margins, haulage quotes, customer commitments
My controllers are going to feel their job is being taken away.
They gain the opposite: their knowledge becomes the raw material of the plan, it carries their names, and it stops going on holiday with them.
What the half-year shows, with figures:
· 41 of the network's 118 constraints came from them, and they were written down nowhere. They are now in the plan, with the name of whoever supplied them and the date they were last verified.
· Three weeks of absence by one of them cost €11,000 last year. That figure says what their knowledge is worth, and it is now written down in a document the board reads.
· 46 hours a month of order keying disappear, and that is the part of their job nobody ever claimed.
What they now do, and never had time for:
· They arbitrate disruptions on three costed answers, in thirty seconds, instead of rebuilding a plan at 5 in the morning.
· They own the simulator: testing an organisation takes two minutes, and their ideas finally reach committee with a figure beside them.
· They run the monthly constraints review — all 118 get checked and enriched, and 14 new ones have come in over six months, all from them.
The figure that settles the question: of the half-year's 214 replans, 214 were decided by a controller. None was applied without a human, and the log shows it line by line.
What I propose: that the monthly constraints review be formally entrusted to them, with the costed effect of every constraint added. The 14 supplied over six months are worth €9,400 of redeliveries avoided — that is a result that belongs in an annual review.
What the half-year shows, with figures:
· 41 of the network's 118 constraints came from them, and they were written down nowhere. They are now in the plan, with the name of whoever supplied them and the date they were last verified.
· Three weeks of absence by one of them cost €11,000 last year. That figure says what their knowledge is worth, and it is now written down in a document the board reads.
· 46 hours a month of order keying disappear, and that is the part of their job nobody ever claimed.
What they now do, and never had time for:
· They arbitrate disruptions on three costed answers, in thirty seconds, instead of rebuilding a plan at 5 in the morning.
· They own the simulator: testing an organisation takes two minutes, and their ideas finally reach committee with a figure beside them.
· They run the monthly constraints review — all 118 get checked and enriched, and 14 new ones have come in over six months, all from them.
The figure that settles the question: of the half-year's 214 replans, 214 were decided by a controller. None was applied without a human, and the log shows it line by line.
What I propose: that the monthly constraints review be formally entrusted to them, with the costed effect of every constraint added. The 14 supplied over six months are worth €9,400 of redeliveries avoided — that is a result that belongs in an annual review.
We have a total transport cost of €6.2M. We do not know where it goes.
It is broken down: by lane, by branch and down to the line delivered. And the gap between your branches runs from €2.14 to €9.80 a line.
How the cost is rebuilt: I match carrier invoices, transport-system orders and ERP order lines, then allocate every euro to the lane, the branch and the line that actually consumed it. 18 months, 1.4 million movements, €6.2M allocated to 99.4 %.
What the breakdown shows:
· 11 lanes carry 74 % of the cost. The other 30 share €1.6M.
· Average cost of a line delivered: €3.42. From €2.14 at the best-served branch to €9.80 at the worst-served.
· 3 branches represent 4 % of volume and 11 % of cost. And for each one the cause is identified: Vaubourg is served twice a week for 1.2 pallets each time; Castevrin sits at the end of a lane, 74 km beyond the last drop; Trémeuse still receives a dedicated run set up in 2019 for a customer that has since closed.
· The Trémeuse run costs €41,000 a year and today serves 3 customers who would be delivered next-day by the existing lane. Nobody had reopened it in four years.
What that changes in steering: you stop discussing a total that says nothing, and you arbitrate on eleven lanes and three branches. Those fourteen lines carry 78 % of what can actually be acted on.
What I propose: I publish this cost every week, by lane, with the gap to the plan retained. And I start with the most absurd item in the network, the one I flagged on day one. real-cost_per-lane-and-branch.pdf€6.2M allocated to 99.4 %, €2.14 to €9.80 a line
⛓ Sourced · carrier invoices (18 months), transport-system orders, ERP order lines
How the cost is rebuilt: I match carrier invoices, transport-system orders and ERP order lines, then allocate every euro to the lane, the branch and the line that actually consumed it. 18 months, 1.4 million movements, €6.2M allocated to 99.4 %.
What the breakdown shows:
· 11 lanes carry 74 % of the cost. The other 30 share €1.6M.
· Average cost of a line delivered: €3.42. From €2.14 at the best-served branch to €9.80 at the worst-served.
· 3 branches represent 4 % of volume and 11 % of cost. And for each one the cause is identified: Vaubourg is served twice a week for 1.2 pallets each time; Castevrin sits at the end of a lane, 74 km beyond the last drop; Trémeuse still receives a dedicated run set up in 2019 for a customer that has since closed.
· The Trémeuse run costs €41,000 a year and today serves 3 customers who would be delivered next-day by the existing lane. Nobody had reopened it in four years.
What that changes in steering: you stop discussing a total that says nothing, and you arbitrate on eleven lanes and three branches. Those fourteen lines carry 78 % of what can actually be acted on.
What I propose: I publish this cost every week, by lane, with the gap to the plan retained. And I start with the most absurd item in the network, the one I flagged on day one. real-cost_per-lane-and-branch.pdf€6.2M allocated to 99.4 %, €2.14 to €9.80 a line
⛓ Sourced · carrier invoices (18 months), transport-system orders, ERP order lines
The crossed flow you mentioned at the start — what is it exactly?
The same part numbers going up and coming down in the same week: 63 tonnes a year, €78,000, and 47 occurrences in eighteen months.
What happens, and why it is invisible: two buyers order the same part, one for North, one for South. When one is long and the other short, operations rebalance with a transfer. Each transfer is justified on its own; it was the matching of the two that was done nowhere.
The detail, measured:
· 47 occurrences in 18 months, on 34 part numbers, all fast-moving.
· 63 tonnes a year crossing, that is €78,000 of transport and 21,000 km driven for nothing.
· In 41 cases out of 47, the two transfers leave less than six days apart.
The correction I propose, written and tested: an allocation rule assigning every fast-moving part to a home centre, and a check that blocks any crossed transfer on the same part within a ten-day window, offering instead to rebalance at source.
What the eighteen-month replay gives: 44 of the 47 occurrences removed, €74,000 a year, and 3 cases kept — three site emergencies where the crossed transfer was the right call. The rule lets those through and flags them, rather than forbidding them.
What it costs to put in place: half a day of configuration in the ERP, and your three centre managers approving the home-centre list. The return is immediate: the first week of application removed two transfers, that is €3,100. crossed-flows_47-occurrences.pdf€74,000 a year, 3 legitimate cases kept
⛓ Sourced · inter-centre movements (18 months), item master, transfer orders
What happens, and why it is invisible: two buyers order the same part, one for North, one for South. When one is long and the other short, operations rebalance with a transfer. Each transfer is justified on its own; it was the matching of the two that was done nowhere.
The detail, measured:
· 47 occurrences in 18 months, on 34 part numbers, all fast-moving.
· 63 tonnes a year crossing, that is €78,000 of transport and 21,000 km driven for nothing.
· In 41 cases out of 47, the two transfers leave less than six days apart.
The correction I propose, written and tested: an allocation rule assigning every fast-moving part to a home centre, and a check that blocks any crossed transfer on the same part within a ten-day window, offering instead to rebalance at source.
What the eighteen-month replay gives: 44 of the 47 occurrences removed, €74,000 a year, and 3 cases kept — three site emergencies where the crossed transfer was the right call. The rule lets those through and flags them, rather than forbidding them.
What it costs to put in place: half a day of configuration in the ERP, and your three centre managers approving the home-centre list. The return is immediate: the first week of application removed two transfers, that is €3,100. crossed-flows_47-occurrences.pdf€74,000 a year, 3 legitimate cases kept
⛓ Sourced · inter-centre movements (18 months), item master, transfer orders
We renegotiate transport in January. What can you bring to that?
A file per carrier, and an 18 % gap you had no way of seeing: on four comparable lanes, one of your five carriers sits above the median of the other four.
What the file holds, carrier by carrier:
· The volume you place with them: from €2.1M to €190,000 a year, and their share of your total cost.
· Their measured service level, not their declared one: promised dates against actual dates across 18 months. From 98.7 % to 91.4 % depending on the carrier.
· Their cost per kilometre on comparable lanes — same distance, same vehicle type, same time window. It is the only comparison that stands up in front of them, and it is built on 4 matched lanes.
· Their disruptions: number, emergency haulage cost attributed, customer commitments missed.
The gap the comparison brings out: the carrier holding the Centre → East lane sits 18 % above the median of the other four, at identical conditions, across 4 matched lanes. On their annual volume of €840,000, aligning to the median is worth €151,000.
The three levers, costed: tariff alignment to the median of comparable lanes (€151,000) · a 97 % service-level commitment with a penalty, which would take their disruption cost from €31,000 to €8,000 · a longer contract against a volume rebate, to be costed in the room.
And the tender is ready, should you want to open the market: 5 lots built on your real lanes, actual volumes attached, service commitments and penalties drafted, weighted scoring grid included. Simply showing it moves a negotiation — and you then decide whether to launch it or not. negotiation-file_5-carriers.pdf18 % measured gap, €151,000 at stake
⛓ Sourced · invoices and contracts of the 5 carriers, promised and actual dates, matched lanes
What the file holds, carrier by carrier:
· The volume you place with them: from €2.1M to €190,000 a year, and their share of your total cost.
· Their measured service level, not their declared one: promised dates against actual dates across 18 months. From 98.7 % to 91.4 % depending on the carrier.
· Their cost per kilometre on comparable lanes — same distance, same vehicle type, same time window. It is the only comparison that stands up in front of them, and it is built on 4 matched lanes.
· Their disruptions: number, emergency haulage cost attributed, customer commitments missed.
The gap the comparison brings out: the carrier holding the Centre → East lane sits 18 % above the median of the other four, at identical conditions, across 4 matched lanes. On their annual volume of €840,000, aligning to the median is worth €151,000.
The three levers, costed: tariff alignment to the median of comparable lanes (€151,000) · a 97 % service-level commitment with a penalty, which would take their disruption cost from €31,000 to €8,000 · a longer contract against a volume rebate, to be costed in the room.
And the tender is ready, should you want to open the market: 5 lots built on your real lanes, actual volumes attached, service commitments and penalties drafted, weighted scoring grid included. Simply showing it moves a negotiation — and you then decide whether to launch it or not. negotiation-file_5-carriers.pdf18 % measured gap, €151,000 at stake
⛓ Sourced · invoices and contracts of the 5 carriers, promised and actual dates, matched lanes
Six months. What does it give, in euros and in service?
€470,000 of annualised saving, and lead-time performance moves from 96.2 % to 98.7 % at the same time. Both together, not one against the other.
The four sources of gain, each tied to its document:
· €74,000 — crossed flows removed, 44 occurrences out of 47.
· €164,000 — reorganisation of the 11 lanes carrying 74 % of the cost, including the two lanes at 51 % fill.
· €78,000 — opening the Saturday dock at Centre, net of the dock's cost.
· €154,000 — carrier negotiation, alignment obtained on two of the three levers.
What service gained at the same time:
· Lead-time performance: 96.2 % → 98.7 %, and the 3 branches at 84.1 % are now at 97.9 %.
· Redeliveries: 210 → 52 a year.
· Capacity overruns: one week in three → 2 weeks in 26.
The time, on your three measured tasks: building a complete plan 70 % → 9 %, checking capacities 40 % → 5 %, assessing margins 30 % → 6 %. A complete plan took 3 weeks; it takes one day, and there are three of them.
The figure that does not flatter me: on two branches, lead-time performance dropped by 1.4 points for six weeks, in February and March.
The cause, once found, was simple: a customer of those two branches had changed its receiving window — it no longer took delivery before 09:00 — and that change came through no channel at all. My constraints were up to date with what you had told me; they were not up to date with what the customer had decided.
What I did with it, and the result: I now read refusals and redeliveries from the transport system automatically, and three refusals on the same customer within a fortnight trigger a check of its window. Detection time moves from six weeks to 48 hours. Over the last ten weeks: 4 customer windows corrected, and 0 lead-time drops. The rule covers all 34 narrow-window customers, not just that one. half-year-review_470000-euros.pdf4 sources of gain, lead times at 98.7 %
⛓ Sourced · carrier invoices, transport system, warehouse system, log of plans retained, redelivery records
The four sources of gain, each tied to its document:
· €74,000 — crossed flows removed, 44 occurrences out of 47.
· €164,000 — reorganisation of the 11 lanes carrying 74 % of the cost, including the two lanes at 51 % fill.
· €78,000 — opening the Saturday dock at Centre, net of the dock's cost.
· €154,000 — carrier negotiation, alignment obtained on two of the three levers.
What service gained at the same time:
· Lead-time performance: 96.2 % → 98.7 %, and the 3 branches at 84.1 % are now at 97.9 %.
· Redeliveries: 210 → 52 a year.
· Capacity overruns: one week in three → 2 weeks in 26.
The time, on your three measured tasks: building a complete plan 70 % → 9 %, checking capacities 40 % → 5 %, assessing margins 30 % → 6 %. A complete plan took 3 weeks; it takes one day, and there are three of them.
The figure that does not flatter me: on two branches, lead-time performance dropped by 1.4 points for six weeks, in February and March.
The cause, once found, was simple: a customer of those two branches had changed its receiving window — it no longer took delivery before 09:00 — and that change came through no channel at all. My constraints were up to date with what you had told me; they were not up to date with what the customer had decided.
What I did with it, and the result: I now read refusals and redeliveries from the transport system automatically, and three refusals on the same customer within a fortnight trigger a check of its window. Detection time moves from six weeks to 48 hours. Over the last ten weeks: 4 customer windows corrected, and 0 lead-time drops. The rule covers all 34 narrow-window customers, not just that one. half-year-review_470000-euros.pdf4 sources of gain, lead times at 98.7 %
⛓ Sourced · carrier invoices, transport system, warehouse system, log of plans retained, redelivery records
What do you do without being asked?
Three gestures, and all three stay inside the plan you retained. Anything that commits means or money lives in a mandate you write and withdraw in one word.
· Replan inside the retained plan when something goes wrong — move parcels onto a lane that has margin, shift a dock slot —, without leaving the 118 constraints or the envelope, and telling you about it. 214 replans this half-year, 0 outside constraints, 0 outside budget, and every one decided by a controller.
· Flag a capacity overrun 48 hours before it happens. 31 flags this half-year, 28 handled before the date. That is what took overruns from one week in three to two weeks in twenty-six.
· Publish the real cost per lane every week, including when it rises. 26 publications, 4 of them reporting a rise — and it was one of those four that uncovered the changed customer window.
What the mandate covered: 238 spot movements in six months, 0 above the €4,500 per-movement cap, 0 outside the envelope, 2 quotes compared every time. Median time from signing the plan to orders leaving: 8 minutes.
What I propose widening, costed: raising the per-movement cap from €4,500 to €7,000 would take movements needing approval from 18 to 4 per half-year. Across those 18, approval delay cost 4 h 20 on average, and twice the best quote was no longer available at signature — €2,900 lost. You keep your hand on the four largest, and you recover the €2,900. framework-of-gestures_and-mandates.pdf214 replans, 0 outside constraints, 0 outside budget
✎ Framework · replanning log, haulage log, weekly cost publications
· Replan inside the retained plan when something goes wrong — move parcels onto a lane that has margin, shift a dock slot —, without leaving the 118 constraints or the envelope, and telling you about it. 214 replans this half-year, 0 outside constraints, 0 outside budget, and every one decided by a controller.
· Flag a capacity overrun 48 hours before it happens. 31 flags this half-year, 28 handled before the date. That is what took overruns from one week in three to two weeks in twenty-six.
· Publish the real cost per lane every week, including when it rises. 26 publications, 4 of them reporting a rise — and it was one of those four that uncovered the changed customer window.
What the mandate covered: 238 spot movements in six months, 0 above the €4,500 per-movement cap, 0 outside the envelope, 2 quotes compared every time. Median time from signing the plan to orders leaving: 8 minutes.
What I propose widening, costed: raising the per-movement cap from €4,500 to €7,000 would take movements needing approval from 18 to 4 per half-year. Across those 18, approval delay cost 4 h 20 on average, and twice the best quote was no longer available at signature — €2,900 lost. You keep your hand on the four largest, and you recover the €2,900. framework-of-gestures_and-mandates.pdf214 replans, 0 outside constraints, 0 outside budget
✎ Framework · replanning log, haulage log, weekly cost publications
One last point: our logistics costs and load plans, where do they go?
They stay with you. I run on local inference on a machine at the company, and your cost per kilometre crosses no outside network to be computed.
Local inference means the model computes on your machine: a load plan or a carrier invoice does not leave your network to be analysed. If you would rather not host a machine, the other route is an isolated resource hosted in France, under French law, dedicated to your network — no pooling with another distributor, and far less with a carrier.
Why this is decisive in negotiation: your cost per kilometre by lane, your fill rates and your capacity margins are exactly what a carrier would pay to know. Knowing that the Centre → East lane runs at 51 % fill, or that three lanes depend on a single provider, means knowing your position before you walk into the room. Those facts stay inside your walls, under French law, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity, and they train no model.
What that looks like day to day: role-based access — rights follow the job: a centre manager opens their site and its lanes, not the carrier contracts or the other sites' costs —, encryption in transit and at rest, and a full log: who retained which plan, changed which constraint, covered which movement, to the second. That log answers your statutory auditor and your insurer in one page.
And you keep control wherever you are: a web dashboard, and supervision from your phone — you approve an above-cap movement or withdraw a mandate in one message, at 4 in the morning as easily as in a meeting.
The next step I propose: the 41 lanes and the 3 centres are covered. Your 22 branches, meanwhile, organise their final delivery rounds each on their own — €3.1M a year, and no shared plan. The route planning agent picks up exactly where I stop, and I already hand it my load plans. I costed the pair on the 4 densest branches: €214,000 a year and 1.2 points more lead-time performance. Say the word and I hand you the full study on Friday. technical-framework_where-your-costs-live.pdfLocal inference, role-based access, processing in the EU targeted
✎ Framework · deployment architecture, role-based access matrix, log of plans and mandates
Local inference means the model computes on your machine: a load plan or a carrier invoice does not leave your network to be analysed. If you would rather not host a machine, the other route is an isolated resource hosted in France, under French law, dedicated to your network — no pooling with another distributor, and far less with a carrier.
Why this is decisive in negotiation: your cost per kilometre by lane, your fill rates and your capacity margins are exactly what a carrier would pay to know. Knowing that the Centre → East lane runs at 51 % fill, or that three lanes depend on a single provider, means knowing your position before you walk into the room. Those facts stay inside your walls, under French law, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity, and they train no model.
What that looks like day to day: role-based access — rights follow the job: a centre manager opens their site and its lanes, not the carrier contracts or the other sites' costs —, encryption in transit and at rest, and a full log: who retained which plan, changed which constraint, covered which movement, to the second. That log answers your statutory auditor and your insurer in one page.
And you keep control wherever you are: a web dashboard, and supervision from your phone — you approve an above-cap movement or withdraw a mandate in one message, at 4 in the morning as easily as in a meeting.
The next step I propose: the 41 lanes and the 3 centres are covered. Your 22 branches, meanwhile, organise their final delivery rounds each on their own — €3.1M a year, and no shared plan. The route planning agent picks up exactly where I stop, and I already hand it my load plans. I costed the pair on the 4 densest branches: €214,000 a year and 1.2 points more lead-time performance. Say the word and I hand you the full study on Friday. technical-framework_where-your-costs-live.pdfLocal inference, role-based access, processing in the EU targeted
✎ Framework · deployment architecture, role-based access matrix, log of plans and mandates
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 agent, several scales of planning. All these uses work in support, subject to your approval.
Included in your agent
The 3 capabilities essential to this promise are included, at no extra cost.
From 623 € excl. VAT / month
Multi-site plans
Builds complete arrangements across your whole network.
Explicit criteria
Optimises cost, lead time or robustness, and states what each plan gives up.
Margins set out
Flags the links whose margin is tightest in each plan.
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 alerts, scenarios and operational trade-offs
What the agent must be connected to
These 2 connections are required for the agent to work. They concern your information system and are scoped during the audit.
Connect ERP/WMS/MES/IoT without creating double entry
Honour constraints, statuses, approvals and error recovery
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.
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 arrangements can a department compare?
By taking on the building of the plans, the effort shifts towards deciding between cost, lead time and robustness. How large the gain is depends on your volume and remains to be confirmed by a pilot.
Building a complete plan
Checking capacity by site
Assessing the margins
Indicative figures, not contractual, to be confirmed by a pilot on your number of sites, links and constraints. Choosing between cost and robustness, and committing resources, are matters of your service policy: those decisions belong to the logistics department.
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 logistics optimisation agent (multi-site plans, criteria, margins), installed and operated for you.
Agility
Setup + controlled subscription
8,755 € excl. VAT setup
then 623 € 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
1,108 € 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
12,560 € excl. VAT setup
then 820 € 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 network
Your logistics costs stay with youLocal inference or an isolated resource hosted in France; no cost and no load plan entrusted to a third party, no data used to train a model.
Data in France, under French lawYour logistics costs and your load plans: minimisation and location in France, architecture designed to reduce exposure to extraterritorial legislation, location alone not being enough to guarantee immunity.
The logistics department keeps the decisionThe agent produces logistics plans that are optimised and compared, which can be checked and altered; no approval is automated.
Human oversight & traceabilityOn your number of sites, links and constraints: systematic logging and tracking, in line with the AI Act.
Going further
Related resources
Frequently asked questions
Your questions, our answers
Does the agent commit resources?
No. It proposes complete plans with their constraints and their margins; chartering, opening a link or adding capacity at a site rest with the logistics department.
How does this differ from round planning?
Round planning organises the day for a set of vehicles; this offer optimises flows between several sites over a longer horizon.
How are the trade-offs presented?
Every plan states the criterion it optimises and what it gives up elsewhere, with the links whose margin is tightest.
What constraints can it take into account?
Capacities by site, promised lead times, transport costs and any constraint you set out at the design stage.
Are our costs protected?
Yes. The agent is 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. Your data is not used to train a third-party model.
How long does it take to deploy this agent?
A few months as a rule, depending on the number of sites, links and how detailed your constraints are, after a free audit then phases of design, integration and testing.
Related use cases
Other agents for logistics
Let's talk
Let's size up the potential in your logistics network
15 minutes to frame your sites and your constraints — hosted in France, supervised, with no commitment.