The same data typed three times
Every handoff, a possible mistake.
Manufacturing · logistics · warehouse
Free pilot for 2 teams — October 2026
Industrial automation and AI for SMEs: fewer manual handoffs, clearer operations.
I take one real workflow — warehouse, production, data, communication — and connect it to the people and systems you already have. Without rebuilding everything.
Nothing stops. Nothing gets rebuilt from scratch.
- ERP
- Excel
- Email and chat
- Scanners and terminals
Integrations · alerts · dashboards
Data entered once, seen by whoever needs it
When the process only holds because people compensate
If your process lives in spreadsheets, chats and people's memory, it isn't a process. It's a dependency.
Work keeps moving because people compensate. They aren't the problem: the system using them as a bridge is.
Information stuck in a chat
One person knows. Everyone else finds out late.
Checks that rely on memory
Works until that person goes on holiday.
The spreadsheet next to the ERP
Started for convenience, now the unofficial truth.
Alerts that arrive late
The problem was visible hours earlier.
Always having to ask someone
If it takes a phone call, the process depends on who picks up.
Small frictions. Every day. They turn into delays and rework.
What I can build
Operational results, not tech for show
I don't replace your ERP, MES or WMS. I work next to them and plug the gaps between systems.
Connect data and tools
APIs, webhooks, files. No copying by hand.
Less double entry
Data goes in once, where it starts.
Visible status and exceptions
What's stuck, what's late, where work piles up.
Cross-team alerts
Requests and approvals with a path, not chats and paper.
Traceability
QR codes, scans, tablet forms: who, what, when.
AI where it helps
Reading documents, routing requests, summarising issues. With a person checking.
Signals from the floor
Arduino, ESP32, sensors. When they're actually needed.
I don't sell hardware by the kilo. I only suggest it when a physical data point is missing and there's no simpler way.
From the floor to the software
Constraints first, technology second
I also work in logistics. You can't judge a workflow from a dashboard: it has to survive shifts, rush orders, absences and exceptions.
And I come from a software company where maintenance and thin resources were normal. I design for whoever uses the tool at 6 a.m. and whoever maintains it two years from now.
- If it slows operators down, nobody will use it.
- Every exception handled by word of mouth is lost data.
- Fewer moving parts beat one more feature.
How we work
Six steps. Stopping is an option too.
No step requires stopping production.
Site visit or call
I look at the real process, not the one in the procedure.
One pain point
The one with the most errors, waiting or repeated questions.
Constraints
Systems, data, security, people, budget.
Prototype
Small, testable, connected to real systems where possible.
Field test
Operators try it. Their opinion beats mine.
Decision
Extend, adjust or stop.
Stopping after a pilot costs less than stopping after a project.
Example interventions
Four typical scenarios
Possible scenarios, not case studies
A.Requests and issues
- Report
- Form
- Assignment
- Alert
- Dashboard
Whoever spots the problem reports it. It reaches whoever must act, and stays tracked.
B.Scattered data
- Files and systems
- Validation
- Single database
- Report
One controlled store, one always up-to-date report.
C.Physical events
- Scan or sensor
- Trigger
- Record
- Alert
The event records itself and alerts the team.
D.Documents
- Document
- Extraction
- Human review
- Approval
AI pre-fills, a person approves.
How much will you save? Your data will tell, not this page.
AI: useful, not magic
AI doesn't fix a broken process
First clear data, responsibilities and handoffs. Then AI, only where it removes real work.
Where it helps
- Routing requests.
- Extracting data from orders, delivery notes, certificates.
- Summarising issues.
- Searching internal docs, with the source.
On what terms
- A person checks before it becomes a decision.
- We decide together which data leaves the company. If needed, models run locally.
- Minimal, documented access.
Who it's for
You need a problem, not a trend
A fit if
- You have a concrete operational problem that keeps coming back.
- Someone can explain how it really works.
- You're willing to try a small pilot.
Skip it if
- You want to replace every system in a week.
- You want "an AI solution" to show off.
- You want to buy technology before understanding the problem.
Freelancer or small agency? This way
Who's behind this
I write software, but I know the warehouse
Timings, handoffs, information that never arrives: I've seen them live, not on a slide.
Full stack developer (Python, Flask, Quart, Docker in production), n8n and Node-RED for automation, Computer Engineering student. I've also written about a Flask app born from real warehouse problems.
- Understand the problem first. Then decide whether to build.
- A good automation keeps working when I'm not around.
FAQ
Questions companies ask
Do we need to replace our ERP, WMS or MES?
No. I work next to them: connecting, collecting data, adding what's missing.
Can we start with a single process?
Yes, that's the right way. One workflow, one result you can check.
Do we need clean data first?
No, but we need to know where it is. Often the first job is cleaning it.
Is AI required?
No. Integrations, rules and alerts are often enough.
Can you work with Excel, email and our existing software?
Yes, that's the most common case. We start there.
Do you do hardware and IoT as well?
Yes, when needed: Arduino, ESP32, sensors. Only if there's no simpler way.
How do you handle data and access?
Only the access needed, agreed in writing and documented. We decide together which data can leave. Any compliance requirements are checked with you before we start.
What does the free pilot include?
2 spots until October 2026. Mapping one workflow, a proposed change and, if it makes sense, a first integration. It's not a full project for free. In return I share the journey as a case study, only with what you approve.
How long does a first test take?
Depends on the workflow and access. The goal is something to try within a few weeks. A precise estimate comes after the mapping.
What happens after the pilot?
You decide: extend with a clear quote, adjust, or stop. No commitment.
Contact
Is one recurring handoff slowing your operations every day? Let's map it and decide whether automation is worth it.
Free pilot for 2 teams — October 2026
Call or site visit. If it's not worth it, I'll tell you. Yes, I know it's a terrible sales technique.