Search
EN
Automation · today's economics

Automation too complex? Not anymore.

Most companies still price automation with 2021 numbers. What never paid off is now viable, with a human in control.

call copy0-1

A little while back, caution was rational. Now it's expensive.

Months-long projects, heavy investment, systems that needed reworking on every process change — that was the calculation of the past. That world is gone: AI hasn't just made processes smarter, it has fundamentally changed their economics. So the real question is no longer "Where can we use AI?".

Which processes0-1

Costs came down. Capability shot up. In 2026 they crossed.

The line you're on didn't bend slowly — it stepped. What a system can reliably take on today is a different order from even a year ago, at a fraction of the cost. Which means the "not worth it" you decided before this year was answered with old numbers.

costs new0
How we work

Discovery still comes first. It just stopped being a workshop.

Keine monatelangen Sitzungen: Wir analysieren eine Woche lang reale E-Mails, Dokumente und Systemprotokolle aus Ihrem Arbeitsalltag.

The honest question is no longer whether it can be done — almost anything can now. It's whether it's worth doing: the cheapest path, and where it breaks without a person. That answer sits in your data, not in a meeting room.

And sometimes the answer is: not yet. We'll tell you that too. We don't sell you an automation — we find out whether it's worth building first.

Start with discovery 40

It almost always comes down to the same three jobs.

This is not about automating single tasks. It is about how you decide, how you serve your customers, and how you run the business.

Docs0

Document Intelligence

Every business runs on documents. Most still have people retyping them.

Invoices, orders, customs documents, forms and emails are read, validated and processed automatically, with a person only where it matters.

 

Manual work0

Operational Execution

Your team should not spend its day copying data between systems.

Digital workers execute repetitive business processes across your applications, while people stay in charge of exceptions and decisions.

 

Legacy0

Legacy Modernization

Yesterday's integration platforms should not hold back tomorrow's business.

Magic xPI, Lobster, and legacy middleware give way to cloud-native Azure integrations and AI-driven workflows, delivered in weeks rather than years.

 

Human in the loop

We built the technology that keeps the human in control.

Agents do the work and check each other. A person authorizes the result, and the system learns from every correction. What we promise does not depend on this year's model:

✔ A person approves, and answers for, every outcome

✔ Every result is verified before anyone sees it

✔ Corrections adjust the process as it drifts

Where it makes sense, this changes more than one process. It changes how decisions get made and how customers are served.

Intervention-20

Not a lab. Our customers' daily work.

For us, AI means in production, not in a demo — in business-critical processes, with a human in control, reliable and scaled in real operation. Measurable value, not an experiment.

  • one to the side 20

     

     

    CLIENT
    Blue Lake Legal
     
    SITUATION
    A Swiss immigration law firm, 2,000 work-permit cases/year across every canton, on an old platform plus manually kept spreadsheets.
     
    WHAT WE IMPLEMENTED
    A human-in-the-loop case platform on Azure. The salary calculations and document checks that open every case now run automatically, and a person approves every outcome. Same workflow, same 24h commitment.

    RESULT
    Over 23% higher accuracy, 25% less workload in the process, capacity freed for higher-value work. Billed per case, not per licence. 

     

     

     

     

  • customs0-2

    CLIENT
    Customs clearance CH/DE/UK 
     
    SITUATION
    500 complex customs declarations a day, on a fragmented stack of ERP, dozens of shared mailboxes, several customs systems, and Excel tools.
     
    WHAT WE IMPLEMENTED
    An AI case platform turns order emails into ERP-ready cases automatically: read, validate, assign customs + statistical codes, consolidate line items, split multi-declaration emails. A human approves every exception, and the system learns.

    RESULT
    40% of declarations arrive by email, those are now handled by the system (rollout ramping). Time per email case: from 8 to just under 4 minutes, roughly halved. 

     

     
     
     
     
     

  • customs copy0

    CLIENT
     Magic xPI Azure
     
    SITUATION
    Runs high-volume integrations for ~120 clients on a legacy platform (Magic xPI): 450+ projects, weekend outages, 1-hour logs, multi-week deployments.
     
    WHAT WE IMPLEMENTED
    An Azure-native architecture replaces each legacy component with a managed service: one Logic App serves many customers (auto-scaling, pay-per-run), API Management instead of a load balancer, Azure Monitor, native connectors, CI/CD. PoC-validated, migration underway.

    RESULT
    450+ projects → 10–30 Logic Apps (up to 45× fewer); deployments from 4–8 weeks to ~30 minutes; no more weekend outages; cost scales with usage.

     

     

     

Not a lab. Our customers' daily work.

For us, AI means in production, not in a demo — in business-critical processes, with a human in control, reliable and scaled in real operation. Measurable value, not an experiment.

CLIENT
Blue Lake Legal
 
SITUATION
A Swiss immigration law firm, 2,000 work-permit cases/year across every canton, on an old platform plus manually kept spreadsheets.
 
WHAT WE IMPLEMENTED
A human-in-the-loop case platform on Azure. The salary calculations and document checks that open every case now run automatically, and a person approves every outcome. Same workflow, same 24h commitment.

RESULT
Over 23% higher accuracy, 25% less workload in the process, capacity freed for higher-value work. Billed per case, not per licence.
 
CLIENT
A Customs clearance company CH/DE 
 
SITUATION
500 complex customs declarations a day, on a fragmented stack of ERP, dozens of shared mailboxes, several customs systems, and Excel tools.
 
WHAT WE IMPLEMENTED
An AI case platform turns order emails into ERP-ready cases automatically: read, validate, assign customs + statistical codes, consolidate line items, split multi-declaration emails. A human approves every exception, and the system learns.

RESULT
40% of declarations arrive by email, those are now handled by the system (rollout ramping).
Time per email case: from 8 to just under 4 minutes, roughly halved.
CLIENT
A warehouse and delivery services company 
 
SITUATION
Runs high-volume integrations for ~120 clients on a legacy platform (Magic xPI): 450+ projects, weekend outages, 1-hour logs, multi-week deployments.
 
WHAT WE IMPLEMENTED
An Azure-native architecture replaces each legacy component with a managed service: one Logic App serves many customers (auto-scaling, pay-per-run), API Management instead of a load balancer, Azure Monitor, native connectors, CI/CD. PoC-validated, migration underway.

RESULT
450+ projects → 10–30 Logic Apps (up to 45× fewer); deployments from 4–8 weeks to ~30 minutes; no more weekend outages; cost scales with usage.

We know what we're doing, and we know the limits.

In a market full of promises, what counts is knowing the limits. These are the people who build your automation.

Valentin

Valentin Cristea

Automation Architect

Designed Nibo and evolved it over twelve years — from IT ticketing to multi-engine automation to AI agents; builds multi-agent architectures for document-heavy processes with human-in-the-loop checkpoints.

"Code is the new low-code."

Octavian

Octavian Sima

Development Manager

Architect for large integration and cloud landscapes — led a Magic xPI → Azure modernization and builds AI-assisted delivery pipelines with multi-agent workflows on production-grade Azure.

"Find the real bottleneck before you automate — AI only lasts on solid architecture and trusted data."

Thomas

Thomas Madsen

Architect & Process Engineer

Finds the automatable core in messy operations — and builds it. From billing automation and automated on/offboarding (with instant lockout via Azure Entra ID) to full content pipelines.

"The devil is always in the detail, but smart, hard work beats it."

Alex M

Alex Mocanu

DevOps & Automation Engineer

Builds the infrastructure that makes automation run reliably — Azure Kubernetes, GitOps and CI/CD from the ground up; shipped a Swiss law firm's salary-calculation automation and onboarding across multiple customers.

"If I do it twice, I automate it."

foundation

Microsoft-grade where it matters, open where it counts.

Book a call now.