Weighing, labelling, scanning, for example food wholesale
The employee weighs, the system takes over the weight and prints the label with the right data, and scanning posts the goods to the order.
Food & logisticsService
Your systems don't talk to each other. Your people do – and that costs time.
AI automation that starts with your workflow. We connect what you already have and let AI handle the routine: emails, quotes, invoices, data reconciliation. Your rules apply throughout.
AI advises, a human decides · Transparency under the EU AI Act
SubjectInput → AI recognises → Human decides → System posts
SheetA-01
Date24 September 2026
DrawingW&K Connect
Starting situation
Does this sound familiar?
An order arrives by email, is typed into the inventory management system and then again into the invoicing software. The same figures three times, three sources of error.
Your inbox is your real order management. When someone is off sick, their cases stand still.
For a quote, an employee pulls together prices from the supplier portal, stock levels from Excel and the customer history from the CRM. From scratch, every time.
-
You have already bought an AI tool. It answers questions, but it does nothing. → That is what the AI employee is for
Logic first, then the interface
Another tool on top rarely solves the problem: it does not know your special cases and who is responsible for what. So your employees build new manual work around the tool, and in the end there is one more workflow, not one fewer.
Automation works when someone has defined the logic beforehand: which email is an order? What happens if the item is not in the catalogue? Who is asked if the price deviates from the standard? We answer these questions with you: process first, software second.
W&K Connect implements AI automation for companies with 5 to 250 employees: connecting systems, automating routine, using AI sensibly, without replacing your existing IT. Before an interface is built, it is clear which inputs there are, which rules apply and who approves. → How we work
Approach
AI automation in four steps: from the workflow to acceptance
-
01
Map the process
We sit next to the employees who handle the workflow today and count clicks, breaks between systems and follow-up questions. Result: one page describing the current workflow, with the time per case.
1–2 sessions
-
02
Define the logic
We record rules, exceptions, approvals and data sources in writing and in your own words. That is the logic document, and you sign it off. This is where it is decided whether the automation will fit later.
1 week
-
03
Connect and automate
We build the interfaces between your systems and use AI where it understands texts, sorts them or makes suggestions. Where money or commitments are involved, a human decides.
2–6 weeks
-
04
Prove
We measure time per case and error rate, before and after. That is the acceptance.
Result
What integrating your systems delivers
-
Interfaces between your systems: inventory management, CRM, accounting, online shop, email and telephony. Documented and maintainable.
-
Automated workflows for recurring cases: sorting the inbox, recording orders, preparing quotes, matching invoices, reconciling data.
-
AI where it has an effect: in understanding emails, PDFs and photos.
-
The logic document your team understands, as the basis for every later extension.
-
Training for your employees instead of a mere handover.
An example from telephony: when a call comes in, the matching case opens and the note goes straight into it. → AI employees: a call opens the case
From our projects, without customer names
Warehouse workflows linked from scale to route, in which nobody retypes data between the steps.
Managed IT for companies with several locations, with monitoring and a fixed point of contact. We operate what we build.
Where such a chain can lead next, as far as sensors and machines: → Predictive maintenance & industrial AI
Examples
Where integration and automation typically start
The industries are only examples.
Sales and inbox, for example trade
Orders arrive as email, PDF or photo, and someone retypes them. After the switch, the AI recognises the orders and creates them. Only unclear cases go to the clerk for review.
Trade & precious metalsService with several locations, for example a care service or a service provider
The IT has grown over the years, across several locations, and nobody has a complete list. Once everything has been recorded, monitoring reports failures before employees notice them.
Incoming invoices from several channels, in any company
Invoices arrive by post, by email and via portals. The AI assigns them to the case and puts deviations forward for approval.
Your process is different? That is exactly where we start. → First call
What we work with
Odoo, Bitrix24, SAP Business One, HubSpot, our own developments; scales, label printers, scanners, telephony, banking, online shops. We use whatever solves your task. Not tied to any vendor.
When AI integration is right for your company
AI integration is right for you if your company has 5 to 250 employees, uses several systems and your employees move data between these systems every day.
Where AI communicates directly with people, it identifies itself as AI (EU AI Act, Art. 50). AI advises, a human decides. If we process personal data on your behalf, we conclude a data processing agreement (DPA) with you under Art. 28 GDPR. → AI transparency
Frequently asked questions about AI integration and process automation
Does this work with our existing software: scales, phone system, DATEV, online shop?
In most cases, connecting to existing software works. We connect anything that has an interface, a file export or database access. Scales, for example, via their protocol, phone systems via API, DATEV via its document exchange, online shops via their interface. If something has none of these, we check it in the first call.
What data does the AI need, and what if we still work with Excel?
The AI needs the data your process already produces today: emails, documents, lists. Excel is no obstacle to getting started. If a workflow depends on a single file, we suggest moving its data into a system with an interface. Then the automation does not depend on that file.
How quickly are first results visible?
For a clearly defined process, first results come as soon as the automation is running: 2–6 weeks after the logic document with rules and approvals has been signed off. We then measure time per case and error rate against the baseline measured beforehand, so that ‘visible’ is a number and not an impression.
What happens after the project: who operates it, who maintains it?
After the project, either your IT operates and maintains the system using our documentation, or we take over operation and maintenance with monitoring and a fixed point of contact.
Do we need our own IT staff for this?
You do not need your own IT staff for an AI integration. What you need is one person who knows the process: for one or two sessions to map it, to sign off the logic document and for feedback during measurement. We take over the operation, or we onboard your point of contact.
How a pilot for one process is structured and what the price depends on: → Pricing and pilot models
Next step
Name the one workflow that costs your team the most time. In the first call, we estimate what can be automated, where you continue to approve yourself and the price range of the first step.
30 minutes, one process from your daily work, no slides. Reply within one working day.
Whether we connect your existing systems or develop something new is decided by your task. Not by our catalogue.
If no system is in charge yet: → CRM and ERP implementation with AI
First call
One workflow is enough to start
Name the one that costs your team the most time. We estimate what can be automated and where you continue to approve yourself.
30 minutes, one process from your daily work, no slides.