Goods receipt without retyping: 90 seconds instead of 20 minutes per delivery note.
Design material flows from goods receipt to goods issue and connect warehouse automation with your SAP processes.
The task starts with the material flow — goods receipt, put-away, picking, goods issue — and with the documents that accompany each of these steps. The document AI reads CMR notes, advice notes and customs papers, checks them against SAP, and flags only the deviation; barcodes and labels are captured without a handheld scanner. What is carried through the warehouse on paper and typed up today runs as a booking through SAP EWM instead.
Capture the material flow: goods receipt, put-away, picking, goods issue
Read documents automatically — CMR, advice notes, customs papers — and check them against SAP
Capture without a handheld scanner: barcode and label are tied to the process, not the device
In the end there's a booking in SAP EWM instead of a stack of paper on the desk. The review step stays with the person — but only where the document AI flags a deviation.
Yard & transport
Entry time from 30 minutes to 1 minute — measured, not estimated.
Connect time slots, carrier coordination, check-in, gate assignment and check-out into one continuous site process.
Today, time slot, check-in, gate assignment and check-out are three systems and a phone call. Merged together, they become one process: the camera reads the license plate, the gate opens, the arrival is posted, the driver gets their instruction — in their own language, without a call. Manually this costs 30 minutes per entry, semi-automated 6, fully automated one.
Time slot, check-in, gate assignment and check-out as ONE process
License plate by camera, gate and arrival posting without a single phone call
Instructions for the driver in their own language, on the terminal or their phone
The difference is measured, not estimated: 30 minutes per entry manually, 6 semi-automated, one fully automated — with the same staff and no second system alongside SAP.
Production
Spot bottlenecks before they cost you throughput.
Connect production planning, machine data, packaging and material supply with ERP and warehouse processes.
Production planning, machine data, packaging and material supply depend on the same resources as the yard: forklifts, drivers, ramps. Where these resources are continuously assessed, forecasts replace reports — capacity, bottleneck, demand. And the system tells you which vehicle or machine will fail tomorrow, while there's still time to repair it today.
Production planning, machine data and material supply in one resource view
Continuously assess forklifts, drivers and ramps instead of reporting after the fact
Forecast instead of report: capacity, bottleneck, demand
Whoever sees the bottleneck one shift earlier reschedules orders instead of losing them. And maintenance learns today what will fail tomorrow — while there's still time to repair it.
SAP implementation & extensions
First use case in 4 weeks, measurable value within the first quarter.
Translate requirements into system concepts, customizing and extensions, and integrate standard modules and add-ons into operations.
Before an agent executes anything, the ERP has to be able to do it: stage 2 of the roadmap checks the existing backend functions and builds what's missing for execution. Extensions are built where the standard ends — directly in EWM, TM and YL, without a data export and without a middleware project. Not a year-long project: start with one use case, value from the first quarter.
Translate the requirement into a system concept, customizing and extension
Check the backend: what can the ERP already do, what's missing for execution?
Extension directly in EWM, TM and YL — no data export, no middleware project
Not a year-long project: the entry point is a use case, the value shows up in the first quarter, and whatever gets built stays extensible within the standard.
Hardware & IoT integration
Cameras and devices post directly into SAP — without a handheld scanner.
Select and procure devices, and synchronize terminals, barriers, scales and warehouses with your SAP landscape.
Terminals, OCR and license-plate cameras, barriers, scales and racking are not run side by side but hung on the same process. The camera detects the damage on arrival, and the complaint is generated automatically — with photo, date and delivery note. The compute power for this sits in your own server room: no cloud subscription, no data transfer, operating costs equal to the electricity bill.
Select and procure devices: terminal, OCR and license-plate camera, barrier, scale
Hang everything on the same process instead of running it side by side
Compute power in your own server room — no cloud subscription, no data transfer
The camera detects the damage on arrival, and the complaint is generated automatically — with photo, date and delivery note. The only operating cost is the electricity bill.
AI process consulting
Autonomy is earned, not assumed — in six stages.
Identify suitable processes, define decision boundaries and trial controlled AI execution starting from the pilot.
First the process is written down and approved, then it's separated: what can be automated by rules and what needs judgment. The first agents run with mandatory approval and full monitoring; their authority only grows once accuracy is proven against KPIs. Process ownership stays with the department — whoever knows the process also decides on the AI.
Write down and approve the process before an agent touches it
Separate: what can be automated by rules, and what needs judgment?
The first agents run with mandatory approval and full monitoring
Autonomy is earned: an agent's authority only grows once its accuracy is proven against KPIs. Process ownership stays with the department — whoever knows the process also decides on the AI.
AI reads, understands and responds — in the plant, in the warehouse and on the road, in any language.
200 supply-chain emails become 20 tasks — in the order they actually matter.
Supplier correspondence
Advice notes, date changes, complaints: the AI reads the supply chain's mail and sorts it by what actually holds up the chain. You see 20 tasks instead of 200 emails.
Document AI in goods receipt
90 seconds instead of 20 minutes per delivery note. CMR, advice notes and customs papers are read, checked against the order, and the deviation is flagged.
Fault report from production
The shift supervisor speaks, the AI writes: machine and order assigned, maintenance informed. No form, no callback, no lost hour.
Driver communication
Time slot, ramp, delay — automatically, in the driver's own language, around the clock. Dispatch no longer has to call and chase.
02 · Ratings
AI evaluates, compares and recommends — across plant, warehouse and transport.
Answers to real supply-chain questions instead of a monthly report.
Queries instead of reports
"Which carriers had poor ETA accuracy this week?" — an instant answer from TM, EWM and the scale. Not a report that only appears at month end.
Delivery capability forecast
Stock, demand and open orders continuously reconciled: the bottleneck shows up on the screen three weeks ahead — not on the day of the customer commitment.
Production plan against material flow
The plan is checked against incoming goods: which batch is missing for which shift — with a rescheduling suggestion before the line stops.
Recommended actions
Not a number, but a recommendation: "Avoid carrier X today — two alternatives with free capacity." The decision stays with the person.
03 · Image recognition
Cameras continuously see what no human can continuously capture — at the gate, on the line and at the pallet.
Gate check-in in 8 seconds instead of 8 minutes.
Automatic gate check-in
8 seconds instead of 8 minutes: the camera reads the license plate, the gate opens, the system books the arrival in SAP EWM.
Quality inspection on the line
The camera checks every cycle, not just a sample. Rejects are removed before they're packed, stored and shipped.
Damage and load securing
Damage detected on arrival, complaint created — with photo, time and delivery note. Load securing is checked before departure.
Labels, barcodes, batches
Pallet, batch and SSCC are read at entry and loading, without a handheld scanner. Traceability happens along the way.
04 · Programming
AI builds workflows itself and executes them in SAP — without an IT ticket and without a second system.
70% of the routines between plant, warehouse and road are fully automatable.
Document flow
CMR, goods-receipt posting, advice notes — handled fully automatically through SAP EWM. No manual typing, no rework.
Feedback from production
Completion confirmation, consumption and batch are posted from the machine to SAP PP — without a terminal at the line and without a collective posting at shift end.
Autonomous yard and route control
Entry, position, ramp, exit and the sequence of routes — coordinated without a radio call and without misunderstanding.
Predictive maintenance
The system detects which machine and which vehicle will fail tomorrow — repair it today instead of standing still tomorrow.
Four capabilities, one supply chain.Read, assess, see, execute — the same four capabilities in
production, the warehouse and on the road.
The assistant sees all running transports, prioritizes on its own, and presents the cases to the dispatcher.
Disponentenleistung +30 bis +50 %
AI operations
Autonomous dispatching
Defined decisions run entirely without intervention; response time drops from minutes to seconds.
Routine: 90–95 % autonom
SAP
Digital twin of the supply chain
A virtual image of the network simulates scenarios in real time before they become reality.
Grundlage für Stufe 5
Warehouse
The deviation shows up at the gate
The camera assesses the load and load securing, text recognition reads the delivery note, the posting goes directly to SAP EWM — with an image attached to the document.
Manufacturing
The line doesn't wait for raw material
An agent reads the order forecast, stock and shift schedule, calls the transport orders itself, and writes in plain sentences why it decided that way.
Transport
Every freight invoice is checked, not every tenth one
The invoice is read, matched against the freight contract and shipment, and only surfaced where it deviates — with the line item in question.
Warehouse
Empties that nobody has to chase by phone anymore
The quantity is detected during unloading and credited to the carrier's account; the difference reports itself, to the driver and in accounting.
SAP
Kundenanfrage trifft SAP-Auftrag
Die E-Mail wird gelesen, eine fehlende Kundennummer gezielt nachgefragt und die Antwort direkt aus dem SAP-Auftrag gebaut — von der Anfrage bis zur Lösung ohne Bearbeiter.
Bearbeitungszeit: von 15 Minuten auf Sekunden
Automatic plant gate check-in
Yard
ANPR recognizes the license plate, matches it against the TMS, and assigns the ramp slot digitally.
The camera at the gate reads the license plate, the system finds the expected delivery, checks the appointment, papers and access — and sends the driver straight to the right ramp without stopping. The gatekeeper turns from data-entry clerk into decision-maker for the exception.
unter 3 Minuten statt 15–25
ETA forecast
Transport
Models calculate the likely arrival from history, traffic, weather and known bottlenecks.
It's not the plan that says when the truck arrives, it's the model. As a result, unplanned delay notifications to customers also drop by 20 to 30 percent.
87 % Treffer, 4 Stunden voraus
Early disruption warning
Transport
Deviations in GPS and ETA are detected two to four hours before escalation.
The most expensive disruption is the one you only hear about at the ramp. Whoever sees it hours earlier can still plan — instead of reacting.
Ungeplante Krisen −40 bis −60 %
Dynamic route optimization
Transport
Routes keep being optimized while under way; new orders, time slots and traffic feed in continuously.
A route isn't a single decision made in the morning, but a sequence of decisions throughout the day. The model recalculates it as soon as the situation changes.
Fahrleistung −10 bis −20 %
AI picking
Warehouse
Optimal pick paths, similar orders batched together — shorter distances, more throughput per shift.
In manual warehouses, up to 60 percent of picking time is pure travel time. That's exactly where the lever is — not in picking faster.
Wege −15 bis −25 %, Durchsatz +10 bis +20 %
Warehouse layout optimization
Warehouse
Slotting by picking frequency and goods movement, accounting for the seasonal shift.
Today's best storage slot is the wrong one in November. A model that anticipates the shift rearranges before the season arrives.
Pickwege −10 bis −20 %
Inventory forecast
Warehouse
Demand forecast per SKU, with order quantities and safety stock calculated dynamically.
Safety stock is bought uncertainty. The better the forecast, the less of it has to sit on the shelf.
Bestand −10 bis −20 %, Servicegrad gleich
Production and logistics in sync
Manufacturing
MES and logistics coupled in real time: production changes trigger transport adjustments immediately.
The special transport on Friday evening is almost always the bill for information that arrived two days too late.
Optimizing individual routes gets you a few percent. Optimizing the structure of the network gets you the rest.
Transportkosten bis −20 %
Assistant for dispatching
AI operations
The assistant sees all running transports, prioritizes on its own, and presents the cases to the dispatcher.
Not another dashboard, but a pre-selection: what do I need to decide now — and what has the system already resolved?
Disponentenleistung +30 bis +50 %
Autonomous dispatching
AI operations
Defined decisions run entirely without intervention; response time drops from minutes to seconds.
Autonomy is not all or nothing. It's a clearly defined corridor in which the system may act — and a boundary at which it brings in a person.
Routine: 90–95 % autonom
Digital twin of the supply chain
SAP
A virtual image of the network simulates scenarios in real time before they become reality.
Before a decision reaches the yard, it has already run once in the digital twin. What fails there costs nothing.
Grundlage für Stufe 5
The deviation shows up at the gate
Warehouse
The camera assesses the load and load securing, text recognition reads the delivery note, the posting goes directly to SAP EWM — with an image attached to the document.
Wareneingang per Sichtkontrolle und Zettel — eine falsch gesicherte oder unvollständige Anlieferung fiel erst in der Fertigung auf. Die Kamera bewertet Ladung und Ladungssicherung, die Texterkennung liest den Lieferschein, die Buchung geht direkt nach SAP EWM — mit Bild am Beleg. Der Nachweis liegt beim Beleg, nicht in einem zweiten Ordner; die Reklamation ist am Tor entschieden, nicht drei Schichten später.
The line doesn't wait for raw material
Manufacturing
An agent reads the order forecast, stock and shift schedule, calls the transport orders itself, and writes in plain sentences why it decided that way.
Der Nachschub an die Linien lief nach Erfahrung. Fehlte ein Vormaterial, stand die Schicht — und niemand konnte danach sagen, warum. Ein Agent liest Auftragsvorschau, Bestand und Schichtplan, ruft die Fahraufträge selbst und schreibt in Sätzen dazu, weshalb er so entschieden hat. Die Versorgung ist eine Entscheidung mit Begründung, keine Gewohnheit; wer sie prüfen will, liest sie nach.
Every freight invoice is checked, not every tenth one
Transport
The invoice is read, matched against the freight contract and shipment, and only surfaced where it deviates — with the line item in question.
Frachtkosten wurden stichprobenweise kontrolliert. Was nicht in der Stichprobe lag, wurde bezahlt. Die Rechnung wird gelesen, gegen Frachtvertrag und Sendung gestellt und nur dort vorgelegt, wo sie abweicht — mit der Zeile, um die es geht. Die Prüfung ist vollständig, die Arbeit weniger: bearbeitet wird die Abweichung, nicht der Stapel.
Empties that nobody has to chase by phone anymore
Warehouse
The quantity is detected during unloading and credited to the carrier's account; the difference reports itself, to the driver and in accounting.
Tauschpaletten und Rollbehälter wurden auf Zetteln geführt. Der Bestand stimmte selten, geklärt wurde per Telefon. Die Menge wird beim Abladen erkannt und dem Konto der Spedition zugeschrieben; die Differenz meldet sich selbst, beim Fahrer und in der Buchhaltung. Das Leergutkonto ist am Abend richtig, Streitfälle haben ein Bild und einen Zeitpunkt.
Kundenanfrage trifft SAP-Auftrag
SAP
Die E-Mail wird gelesen, eine fehlende Kundennummer gezielt nachgefragt und die Antwort direkt aus dem SAP-Auftrag gebaut — von der Anfrage bis zur Lösung ohne Bearbeiter.
Eine Anfrage zum Lieferstatus kam per Mail, ohne Kundennummer. Bisher hieß das: Postfach, Rückfrage, Auftrag im SAP suchen, Antwort tippen — rund 15 Minuten Bearbeitung. Die KI liest die Mail, erkennt die fehlende Angabe, fragt genau danach, gleicht die Antwort mit dem SAP-Auftragsstatus ab und schickt die fertige Antwort. Aus 15 Minuten werden Sekunden, der Vorgang ist geschlossen, bevor ein Mitarbeiter ihn geöffnet hätte.
Bearbeitungszeit: von 15 Minuten auf Sekunden
Production
Logistics
Transport
From operations, not from a slide.Production, logistics, transport — one chain. Every case carries
its metric and its source, so you can verify it instead of taking it on faith.
The book looks at the shift from a visible to an acting system, based on practice in the yard, warehouse and transport.
From visibility to actionAI meets physical logisticsHuman judgment & responsibility
Markus J. Kaiser
Founder & Managing Director · INTECIO GmbH / intec.ai GmbH
Technology changes. Curiosity stays.
Markus J. Kaiser connects SAP, logistics and artificial intelligence: with INTECIO he brings intelligent systems into physical supply chains, with intec.ai he examines AI-native company building from the ground up.
🚨 NOTE: This LinkedIn post was created with AI and published at my instruction via …
…browser control and published. The image was also created with AI. Operating websites via AI? Getting started is now surprisingly easy. You describe the task – the AI can read in the browser, …
BCS PROJEKTRON ERP. 10 YEARS IN USE, SHUT DOWN IN 3 DAYS BY INTEC.AI.
After ten years of using BCS PROJEKTRON ERP within Intecio GmbH, we're migrating around 10 GB of company data into the AI Engine from intec.ai. Transferring and taking over the processes took three days. After …
Masterclass · Berlin · 16. September 2026 · 16:00–16:55
What happens if we completely rethink a company today – no longer built around fixed software, dashboards, and standardized workflows, but around artificial intelligence?
With intec.ai GmbH, the first German company without staff, I'm pursuing exactly this question. My thesis is radically simple: an AI-native company no longer needs an employee for every task, no longer needs a separate application for every process, and no longer needs a dashboard for every decision. AI becomes the operational control layer of the company. It understands goals, makes decisions, and carries out processes itself within clearly defined limits.
This is fundamentally changing the role of enterprise software too. Instead of rigid applications with fixed interfaces, a flexible system landscape emerges that continuously adapts to the needs of users and companies. Business users increasingly describe directly what they want to achieve, while AI derives functions, workflows, and applications from that.
My masterclass shows, in practical terms and using examples from logistics among others, what the path from an ERP-shaped company to an AI-native company can look like, and what technical, organizational, and economic potential arises from it.
So in the end, the question isn't really whether companies will use AI. What matters is how long classic organizations stay competitive once AI-native companies achieve the same value creation with far fewer people, higher speed, and a completely new cost structure.
Not the project — the first live process. Measurable value in the first quarter.
0 middlewareNative in SAP
EWM, TM and Yard Logistics driven directly. No second platform to maintain alongside it.
15 yearsFrom practice
Every use case comes from a real go-live — with hardware at the gate, not from a slide.
24/7Operations included
Whoever builds the process keeps it running: one point of contact for software, hardware and SAP.
The Numbers
One site, 120 truck movements a day.
3 minutes of phone calls, follow-up questions and re-entry per movement today.
Fully loaded cost of a full-time yard-office employee: €60,000 a year.
6 hCoordination during the day120 movements × 3 minutes — manual work today.
0.8 FTESaved per siteAt 70% of movements handled automatically: 4.2 h a day.
≈ €48kPer year and site0.8 full-time employees at fully loaded cost — before standing and waiting time.
< 12 monthsPaybackOne use case, one site. Every further site pays off too.
Sample calculation based on the stated assumptions — not a commitment. We calculate your figures with your own volumes during the assessment.
Production, logistics, transport — one chain.The hours are spent at the handovers: registration, weighing, documentation. That is exactly where automation pays off.
Founder & Managing Director · INTECIO GmbH / intec.ai GmbH
Markus J. Kaiser connects SAP, logistics and artificial intelligence: with INTECIO he brings intelligent systems into physical supply chains, with intec.ai he examines AI-native company building from the ground up.
SAP and other SAP products named here are trademarks of SAP SE.
Images: Shutterstock / Adobe Stock
This site sets no cookies and tracks nothing. Only the talking face establishes a connection to its provider — your IP address is processed there. Your choice is remembered locally by the browser.
Yard
SAP
Production
Transport
07.09.2026
· Impulse
06.09.2026
· AI in Practice
05.09.2026
· Company
To the program
