Modernise old software without losing the knowledge inside it.
The Access database, the Excel tool full of macros, the in-house software from years ago: they contain rules that are written down nowhere. We use AI to extract them from code, data and user interface, have your specialists confirm them and rebuild the application on a modern foundation. The old software keeps running until the new one demonstrably does the same.
only one colleague still knows it
Your specialists confirm every rule
cross-checked with real past cases
How do you modernise old software with AI?
Understand first, then build. With old software, often called legacy software, AI reads the source code, database, forms and reports of the old application and describes what it does: which data it stores, which rules it calculates, which workflows it covers. Your specialists check this description. The new application is built from it. Finally, old and new process the same real cases, and the switch only happens once the results match.
What AI does not take off your hands: deciding which old rules still apply today. We clarify that with your specialists before anything is rebuilt.
The software works. The question is for how much longer.
In operations
Only one person still knows how it works.
A colleague built the application twenty years ago and has maintained it ever since. He retires in two years. There is no documentation.
In IT
Every update is a risk.
The software needs an old version of Office or Windows. The vendor no longer exists, and neither do security updates.
Day to day
Everything around it is typed in by hand.
The legacy system has no interface. Data moves via Excel export into the ERP, the online shop and back again.
Buying something new often does not fit, because the old software covers exactly your special cases. Reprogramming used to be expensive, because just understanding the old code took weeks. That is exactly the work AI speeds up.
An old costing tool. Your rules in black and white.
A company has been costing its quotations with an Access database since 2006. AI reads the queries, forms and macros and produces a list of rules. Each rule points to where it sits in the old code.
“The database works out discounts, surcharges and small-quantity charges itself. Nobody knows exactly how any more.”
Small-quantity surcharge of 8 % below 50 units, only for customer group B.Costing macro, function “calculate price”
Material surcharge for stainless steel from the materials table, last updated in 2019.Query “material prices”
The express surcharge is hard-coded, with an amount from 2015.Question for sales: does it still apply?
Customer group D has not appeared in any order since 2017.Drop it or keep it?
Your sales team confirms the list. Only then is anything rebuilt, and every rule becomes a test case.
Example with sample data – not a customer reference and no guarantee of results.
From legacy system to new application in four steps.
It starts with a free introductory meeting. After that, you decide after every step whether to continue. The old software stays in operation until the new one demonstrably does the same.
01
Take stock
We review code, database, user interfaces and interfaces to other systems. AI turns this into an overview: what exists, what is connected and what nobody has used for years.
02
Write down and check the rules
AI describes workflows and calculation rules in plain language. Your specialists confirm, delete or change them. After that, you know what the new development will cost.
03
Rebuild
On a modern, maintainable foundation, with interfaces to ERP, CRM or online shop and AI where it takes work off your hands. The checked rule list becomes automated tests. Code and documentation belong to you.
04
Cross-check and switch over
Old and new process the same real cases. Differences are clarified, data is migrated, then the switch happens, area by area if you wish.
What AI does here and what it does not.
What AI does
Read, describe, prepare tests
- Read and explain old code, including languages hardly anyone writes today, such as VBA, Delphi or old PHP.
- Make connections between tables, queries and forms visible.
- Derive test cases from the rules that compare old and new.
- Draft a large part of the new code. We review it and take responsibility for it.
What AI does not do
Decide what still applies today
- Find knowledge that only exists in people’s heads and was never programmed.
- Decide whether an old rule is still right today. Your specialists decide that.
- Guarantee error-free code. That is why we test the new application against the legacy system.
Before we start
Clarify rights, protect data
We modernise software that belongs to you or was developed for you. For purchased software, we first clarify what the licence agreement allows. Often it is then enough to examine workflows, user interfaces and data instead of the program code. Code and data only go to external AI services under a written agreement.
Modernisation pays off when …
… the software is essential to your business
For example for costing, order processing, inspection reports or warehousing.
… only a few people still understand it
And that knowledge would retire along with them.
… technology or security reach their limits
Old Office or Windows versions, no simultaneous working, no interfaces, no access on the move.
And if not: if standard software covers your workflows without detours, it is usually cheaper. We will tell you that in the introductory meeting too.
Advice and delivery from one source.
Whoever analyses your old software also builds the new one.
Measurable
Tested against your legacy system
Old and new process the same real cases. The switch happens when the results match, not after a demo.
Qualification
Certified in AI automation
Certified AI Automation Manager, combined with more than 20 years of software development.
Experience
Software and interfaces since 2002
We know the technology being replaced today from our own development work, and we build with what replaces it.
Frequently asked questions about software modernisation.
Why not simply buy standard software?
If there is one that covers your workflows without detours, that is usually the better route. But old in-house software often covers exactly the special cases that make your business what it is. Then a custom application is the shorter route. We compare both before you decide.
What makes it cheaper with AI?
Understanding the old code, writing documentation and tests: this used to take up a large share of the time. AI now does a good part of this work, and we check the result. How much difference that makes in your case depends on the size and condition of the old software. The inventory shows it.
Are you even allowed to examine our old software?
Your own software or software developed for you: yes, if you hold the rights to it. For purchased software, we check the licence agreement first. As a rule, an authorised user may observe and test how the software works. Decompiling the program code, by contrast, is only permitted within narrow limits. If in doubt, ask your legal adviser.
Will our code and data go to external AI services?
Not without an explicit agreement. We set out in writing before we start where the AI runs and what it may see. For the analysis, test data is often enough instead of real customer data.
Do we have to work twice during the switchover?
No. Your team works with the old software until the new one demonstrably does the same. We migrate the legacy data and check the migration. The switch happens on an agreed date, area by area if you wish.
What does it cost?
That depends on the size and condition of the old software. In the free introductory meeting, you describe what the application does. You then receive a written quote for the inventory. You know what the new development will cost after the inventory, before it starts. We show development and operation separately.
Which software in your company does nobody dare to touch any more?
Tell us in 30 minutes what the application does, what it runs on and who still understands it today. You will find out whether modernisation is worthwhile and what the first step looks like. Free of charge and without obligation.