'/><rect x='255' y='178' width='346' height='346' fill='%23c3a3ff' transform='rotate(-8 428 351)'/><rect x='470' y='156' width='392' height='392' fill='%23a97dff' transform='rotate(6 666 352)'/><rect x='685' y='134' width='438' height='438' fill='%238f4dff' transform='rotate(-4 904 353)'/><rect x='900' y='112' width='484' height='484' fill='%237a2bff' transform='rotate(10 1142 354)'/><rect x='1115' y='90' width='530' height='530' fill='%236800ff' transform='rotate(45 1380 355)'/></svg>)
Complex business processes, automated end-to-end.
From business case to live operations.
Management reporting: the weekly Excel build is gone, the dashboard updates itself
Every week the bank's capital markets division sends leadership a report on revenue per sales unit and product group, actuals against plan. One colleague built it by hand in an Excel file with around 20 cross-references. Half an hour a week, and two to three hours a week for three months around every year-end rollover. When he was on vacation, the report simply did not appear, and corrected PDF versions got lost in inboxes.
Lunatec moved the report into Power BI. Power Query pulls straight from the SQL databases behind the legacy system, DAX measures reproduce every calculation that used to live in the workbook, and the special cases sit in mapping tables the business maintains itself. The dashboard refreshes every day, vacation or not. In the new detail view, managers filter by month, product and top customer without asking anyone.
Read the full case study →
The pilot ran.
It never made it into production.
Before the automotive supplier saved 708,798 euros, its first attempt had already failed: unstable bots, constant outages, more than 300 processes with no prioritization whatsoever, too few trained developers in-house, plus pressure from management for measurable results inside twelve months. That starting position is the normal case.
Classic RPA clicks and types. The moment a case calls for a decision, the bot stops and the file is back on someone's desk.
Of 300 assessed processes, 51 carried any potential at all, and the lion's share of the savings came out of a single area, finance.
The technology is rarely the reason. Operations, exception handling and ownership after go-live are.
Six automations, six solutions
The weekly report hung on one Excel file and one colleague
5,000 installations migrated, and there was no interface to migrate them with
From failed bot experiments to 708,798 euros saved in the first year
1.7 million transactions at a 97.5 percent success rate
Customer service: order turnaround cut from 39 days to 10
Annual financial statements in 15 minutes instead of one to two hours
AI agents decide. RPA bots execute.
Shown here on broker correspondence at an insurer, 1.5 million emails a year.
Intelligent OCR and natural language processing read every incoming broker email including its attachments, whatever format it arrives in.
The agent reads the content in context, identifies the case type and decides what needs to happen. When it needs an instruction, it puts a specific question to a person.
RPA bots pull the required documents, update the connected systems and send a confirmation as soon as the case is closed.
The process runs under service level agreements and regulatory requirements. 98 percent of all cases get through without anyone touching them.
Eight industries we automate
Legal & Tax
saved in the beA legal inbox. Plus annual financial statements in 15 minutes rather than 90.
Public Services
Parking permits, data migration, portal retrieval. On-premise deployment possible.
Manufacturing
Automotive supplier, break-even in month 7. Plus 520 working days saved on order confirmations.
Healthcare
Asklepios Kliniken, 120+ processes, 97.5% success rate.
Energy, Infrastructure, TelCo
M-net, 190,000 transactions a year.
Banking
Relationship management, P&L reporting, Schuldschein loans.
Insurance
98% fully automated, −91% cost per transaction. Six processes at one group.
Construction & Engineering
400 tenders a month, plus bills of quantities straight out of the CAD drawing.
Assess, prioritize, build, operate
What we use along the way
At the environmental agency, the first case came out of a workshop rather than a requirements document.
The technology follows the process. Where the rules are unambiguous, RPA is the better choice over an AI model.
At M-net, career changers from the customer service team were trained up as citizen developers.
The automotive supplier had already bought RPA. Unstable bots, constant outages, nothing to show for it. What was missing was the assessment of 300 processes and the call on which 51 would carry any weight at all.
Implementation is part of the job, and the responsibility does not end at go-live. At Asklepios, 120 processes run at success rates of up to 97.5 percent. Nobody who walks away after the analysis gets to that number.
We work on the process, not on the ticket. At the citizens' office the whole thing took twelve person-days, because someone had first worked out why there is no interface between the online form and the department's case system.
Which process suits RPA, which one document processing, and which one needs an agent. With a break-even calculation and examples from real projects.
What test automation saves, why 92% of SAP migrations finish late, and what the first 90 days of automation look like.
One email a month.
Give us 30 minutes and we'll work out what is possible
Which processes cost you the most time, the most money or the most patience today? We look at them together and answer your questions about the technology, our method and Lunatec itself.