AI for dispatch, planning, warehouse and transport

Connect your systems. Get more done with AI.

AgentMotion Logistik connects your ERP, WMS, TMS and exports into one picture. On top of it, AI works for you: solving disruptions, planning the week, analysing rates, costs and margins – and finding opportunities that get lost in the daily rush. Your team makes the call.

How it works Free demo
Works with your systems
  • ERP
  • WMS
  • TMS
  • Routing
  • CSV exports
  • Excel
  • Rates & contracts
Problem spotted07:44 · forklift breakdown
Week planned97% sure · €0 extra
ApprovedTruck 2 is on its way

Illustration · shown with our synthetic demo companies

ERPOrders & promises
WMSStock & reservations
TMSTrucks, drivers & routes
CRMCustomers & priorities
RoutingDrive times & roads
RatesContracts & penalties
Case DAY-U4

Aster · 6 pallets

  • Order & promise
  • Stock & reservation
  • Truck, driver & dock
  • Terms & risk
One case

01Connect

All your systems. One shared picture.

Orders in the ERP, stock in the WMS, trucks in the TMS, prices in contracts, plans in exports or Excel: AgentMotion Logistik reads them all through adapters and brings them together in one picture. Your systems stay as they are.

  • Adapters and exports, not a migration
  • One order, one case – across every system
  • Missing data is flagged, never guessed

What the AI takes on

From the 07:44 breakdown to the November peak.

On your connected data, the AI takes on work that today lives in spreadsheets, phone calls and five systems – from minutes to months. Your team decides.

Within minutes

Spot disruptionsforklift down, truck late Test fixesevery lever in the digital twin Prepare decisionswith cost, risk and a deadline Draft messagesto customers and carriers

For the week

Plan the weekcertainty against cost Watch deadlinesbefore a lever gets pricier Move peoplehours between areas Absence & vacationdoes one more vacation fit?

For the months ahead

Judge rates fairlymix, team, downtime, pace Money per clientwho earns, who costs Prepare the peaktemps with an order-by date Improve processesnew ideas in a fair tournament
And any time: questions in your own words – answered with figures and sources
Problem brief07:46

6 pallets for Aster miss the 08:45–09:30 window

Cause
Forklift breakdown 07:44
Current plan
No feasible delivery
At stake
€300

02See it coming

Spot problems before the customer calls.

After every change, the digital twin replans your whole day – trucks, drivers, docks, forklifts, breaks and receiving windows. Orders at risk stand out at once, each with a brief in plain numbers.

  • Breakdowns, delays, urgent orders, closures …
  • Cause and money at stake at a glance
  • Early warning, not alarm noise

AI investigator tests 4 levers

  • Deliver from warehouse Bneeds permission
  • Courier vannot available
  • New receiving windowcustomer must agree
  • Expedited forklift repair€22 · saves €1,298

03Solve

Every fix tested before you see one.

An AI investigator works through the levers your company actually has – in the twin, with costs and the effect on every other order. You get the fix that works, with its risks spelled out.

  • Warehouse B, courier, new window, expedited repair …
  • Net saving on your own terms
  • No order is made worse silently
Fix readyby 07:55

Expedited forklift repair: €22

6 of 6 pallets on time · no other order made worse

ApproveReject
Approved 07:52

Delivered 09:25 · on time

04Decide & act

One click – with the facts in front of you.

Every answer becomes a clear decision with a deadline. Once you approve, exactly that plan is carried out and confirmed in the source system. Without approval, nothing happens.

  • Every fix has a deadline: approve by 07:55
  • Reject with a reason – the AI corrects once
  • Read-back confirms it landed
Plan the week · W39as it stands 90%
97%Best value€040 h movedRecommended
100%The most we can€1,71740 h moved24 h overtime4 on Saturday
Sun 20:00no extra cost Mon 20:00+€135 Tue 20:00+€907 Thu 20:00only 90%
Adopt this planapplies from Monday morningAdopted

05Plan the week

How sure are you to clear the week – and what does more certainty cost?

The AI works through many possible ways your week could go – with your volumes, rates and absence. You see how sure each area is to finish by Saturday, which levers improve that, and by when you must decide before it gets more expensive.

  • Moving hours, overtime, temps, Saturday
  • Deadlines: until when it costs nothing extra
  • Adopted in one click
Rate · Returns+6% against fair target
Money per clientOverall margin 16%
Nordwind Outdoor26%
Atelier Nord23%
Lumen Mode12%
Kiezschuh−8%
Kiezschuh: 75% shoes – losing money

06The numbers behind it

What's behind a rate – and where money is lost.

The AI takes every rate apart into mix, team, downtime and pace, and sets a fair target instead of a flat one. It shows what one more unit really costs with each lever – and, with client lines and prices, the margin per client.

  • Fair targets, not flat ones
  • True cost per lever and unit
  • Team level, never per person
Absence aheadsick + vacation per week
One more vacation?Tue1 personFine: no change to the week
Peak season17 temps at the peak

order by 26 Oct!

07Look ahead

Absence, vacation and peak season – before they hit.

How many people will be away next week, area by area? Does one more vacation on Tuesday still fit? How many temps do you need for Black Week – and by when must you order them? The AI works it out from your own history.

  • Absence per week, with a forecast
  • A vacation request checked in seconds
  • Temps with an order-by date

What's the cheapest safe plan for next week?

Move 8 h a day from Support to Returns, Monday to Friday: 97% at no extra cost. 100% costs €1,717 (24 h overtime, 4 people on Saturday). Decide the first lever by Sunday 20:00.

Company data · Week plan (next week)

08Ask

Ask in your own words.

The AI reads the same company data you do – week plan, rates, absence, clients – and answers with figures and their source. It only reads, changes nothing and sees only what you're allowed to see.

  • Answers with their source
  • Read-only, with your permissions
  • Deciding stays with you

Tournament · 7 fresh simulated daysexample

Today
New idea

Only if it wins: adopted after your approval

09Improve

Every decision makes the next one better.

Every case goes into the decision library. In the improvement lab, new ideas compete with your current process on fresh simulated days. Only what wins – and what you approve – goes live.

  • Scorecard for decision quality
  • Fair tournaments with a protected referee
  • You decide what gets adopted

Inside the product

What it looks like in the app.

Shown with our synthetic demo companies: a fulfilment centre and SpreeKern.

Case DAY-U4 · Aster

Problem brief

6 pallets for Aster cannot be delivered in the 08:45–09:30 receiving window.

Open6 of 6 pallets
Receiving window08:45–09:30
Current planNo feasible delivery
At stake now€300.00

What happened

07:44 Forklift breakdown during handling

Customer terms

Standard customer: €30 per started hour late (max €180), €300 if not delivered

Levers the company has

  • New receiving window with the customer€15.00
  • Deliver from warehouse B permission pending€30.00
  • Courier van not available€300.00
  • Move a break€0.00

Schedule estimate from current source records – no forecast.

Why AgentMotion Logistik

Today vs. with AgentMotion Logistik

Instead of five systems, the phone and a spreadsheet: one place for every exception and every plan – from the first signal to the confirmed fix.

TaskTodayWith AgentMotion Logistik
Notice the problemThe customer calls, five systems to checkSpotted early in the digital twin
Find a fixExperience, phone calls, spreadsheetsEvery option tested in the twin
Weigh the costsA rough guessTerms and penalties priced in
Carry it outBy hand, in several systemsApproved plan, confirmed in the source system
Prove itEmails and memoryA complete record of every step
Get betterWhen there's timeScorecard and improvement lab
Plan the weekGut feeling and a spreadsheet formulaCertainty and cost per plan, with deadlines
Judge a rateOne flat target for every dayA fair target from mix, team and downtime
Prepare for the peakOrder temps when it gets tightNeed and order-by date weeks ahead

Features

Everything a good decision needs.

Labelled honestly: Available is in the product today, New is rolling out, In development is on the roadmap.

Solve disruptions

Available

Digital twin

Warehouses, stock, docks, trucks and customers – as the source records describe them. Play, pause, jump back.

New

Autopilot

Starts investigations on its own – biggest money at stake first, within a daily budget. Approving stays with you.

Available

AI investigator

Works through bounded tools, with no access to your systems.

Available

Problem briefs

Pallets, promise, window, cause and money at stake – from source records.

Available

Order desk

Promise only what the whole operation can deliver.

Available

Draft messages

Messages to customers and carriers – prepared, never sent without you.

Available

Permissions per lever

Overtime, warehouse B, a courier: each lever with its own approval.

Available

Audit ledger

Every step traceable – and replayable against today's rules.

New

Decision library & scorecard

Every problem, decision and outcome – and how good the decisions were.

New

Improvement lab

AI agents propose improvements, tournaments show what really helps.

Plan and look ahead

New

Week plan with certainty and cost

The cheapest plan for each level: moving hours, overtime, temps, Saturday – counted at what usually comes through.

Available

Data from any source

ERP, WMS, TMS, exports or Excel – you see which data points were understood.

New

Look-back & proof

What each week cost, where money leaked – and what planning saved, measured on what actually happened.

New

Decision deadlines

By when each lever costs nothing extra – and what waiting costs.

New

Fair rates

Every rate taken apart into mix, team, downtime and pace – team level, never per person.

New

True cost per lever

What one more unit costs: own hour, overtime, temp, Saturday, borrowed hour.

New

Money per client

Revenue, labour and margin per client – recalculated with other prices.

New

Absence & vacation

How many are away next week, area by area – and whether one more vacation fits.

New

Peak season

Temps for the peak, with an order-by date and the extra cost.

New

Assistant for your company data

Questions in your words, answers with figures and their source. Read-only, with your permissions.

Does one more vacation fit in hall 2 next Tuesday?

In developmentComing next

  • Live event intake
  • Exports from SAP, Dynamics, PTV and ATOSS
  • Shadow mode on your cases
  • Morning brief by email or Teams
  • Learned fix ranking
  • Earned autonomy & kill switch
  • A setup form for your costs and penalties
  • Work outside your systems made visible
  • Plans back into your systems and spreadsheets

Security & control

Control your operation can count on.

The engine owns the facts, the AI helps weigh the options, your team makes the decision.

AI proposes, people decide

The AI checks, compares and drafts – it cannot approve, execute or change your systems.

No invented facts

Figures come from your systems, exports and contracts. Missing data is shown as missing.

Approve exactly this plan

Every approval is bound to the checked plan. If anything changes, it's checked again.

A complete record

What was known, what was considered, who decided – and what happened.

Your own deployment

Next to your systems. Your data is never mixed with other customers' data.

Insight, not surveillance

Reports describe processes, never individual employees – with works councils and GDPR in mind.

Evidence

Measured, not claimed.

We test on complete simulated operating days and weeks of our synthetic companies – with the AI and without it, on the same days.

€2,349

lower cost on three AI-managed days than on the same days left alone

66%

of AI answers matched the best available fix – or rightly said there was none (29 cases)

To the cent

our referee reproduces the day results – every measured difference is real

556

candidate fixes measured, from 50 problem cases

18%our forecast22%the site's own

Daily error on new units, estimated before the week: our forecast was closer than the site's own on 108 of 180 area-days (synthetic fulfilment centre).

16 → 1

false “week at risk” warnings in three simulated weeks, after we checked and fixed our own forecast

And we show this too:

  • 3 of 11 executed fixes didn't deliver. One cause is found; the improvement is running in the lab.
  • A realism change made the same days €3,750 more expensive. The lab flagged it instead of quietly moving the yardstick.
  • There isn't enough data yet for a learned fix ranking. It ships when the data supports it.

Simulated operating days of SpreeKern, a synthetic Berlin logistics company, and weeks of a synthetic fulfilment centre (September 2026). Indicative, not statistical, and not customer results.

For the whole team

Site, dispatch and finance – everyone gains.

Dispatch

  • A tested fix instead of hunting through five systems
  • Deadline and consequences at a glance
  • Approve in one click

Site & operations leads

  • The week priced, with deadlines
  • Open problems ranked by risk
  • Fair targets instead of flat ones

Finance & controlling

  • Cost and penalty exposure per decision
  • Margin per client and true cost per lever
  • Outcome against forecast

How we start

Start with what you already have.

No big-bang integration. Prove the value on your own data first, then connect and go live step by step.

  1. 1

    Your data

    exports · no integration

    You give us exports from your systems. On your own numbers, we show which work and which problems the AI can take on.

  2. 2

    Shadow mode

    live cases · nothing executes

    The system plans and recommends alongside your team. You compare – on your own numbers.

  3. 3

    Live with approvals

    your team approves every fix

    Plans and fixes reach your team with deadlines. Approved ones are executed and confirmed.

  4. 4

    Connect live

    ERP, WMS, TMS · live events

    Connected to your systems, the AI solves the day's exceptions too. Where the evidence supports it, low-risk fixes can be pre-approved.

Contact

Free demo.

In 30 minutes we'll show you which work and which problems the AI can take on for you – on our demo companies or on your own data.

  • A product walkthrough on our demo companies
  • An honest view on whether it fits
  • A plan for starting on your data

Your details go straight to our team and are only used to reply to you. Privacy policy

FAQ

Frequently asked questions

Answers on features, security and getting started.

Does the AI act on its own?

No. The AI investigates, compares and drafts. A person approves every action – bound to exactly the checked plan. The autopilot only starts investigations. Pre-approved reversible fixes are on the roadmap and will only ever run within your limits.

Do we have to replace our ERP, WMS or TMS?

No. Your systems stay the source of truth. We read through adapters, write back only approved actions and then read them back to confirm they landed.

Which systems can you connect?

Today: ERPNext, Odoo, a TMS interface, routing estimates and CSV and Excel exports – shown end to end on our demo companies. We build adapters per system family, not per customer. Tell us what you run, and we'll tell you honestly what it takes.

Do we need an integration to start?

No. We start with exports from your systems – CSV, Excel or whatever you have – as they are. You see which data points were understood. Then we connect your systems live, step by step.

Our data is messy. Is that a problem?

It's expected. Valid rows survive, broken rows are quarantined and reported, and missing data is shown as missing. A case without critical facts goes to a person instead of being guessed.

How do you know a fix will work?

Every candidate fix runs through the digital twin and is compared with changing nothing – deliveries, costs and the effect on every other order. After execution, the result is read back from your systems.

Which figures are assumptions?

Volumes, hours, rates and absence come from your data. Hourly costs, premiums and penalties are stated default values until the setup form – every figure says what it rests on. “Certainty” means the share of simulated futures in which every area clears by Saturday.

What happens when the AI is wrong?

Your team rejects with a reason and the AI gets one correction turn. Case, reason and outcome go into the decision library, where the scorecard and the improvement lab learn from them.

Where does it run, and who sees our data?

Each customer gets a dedicated deployment next to their systems. Mappings, rates, cases and outcomes stay yours and are never pooled with other customers' data.

Which AI models do you use?

Leading frontier models through a bounded tool interface – with no credentials for your systems. The engine checks every answer the same way, whichever model wrote it.

What about works councils and the EU AI Act?

Reports describe processes, never individual employees. People can understand, override and stop the system at any time. We don't promise legal compliance across the board – each deployment gets its own assessment.

Are the demo companies real?

No. SpreeKern (logistics in Berlin), a Berlin distributor running its own tours and a fulfilment centre are synthetic companies we use to test the product end to end. Synthetic results prove the mechanism – only your data proves the value for your business.