Duplicates are found and merged, spellings and units made consistent. Analyses no longer start with manual work.
ERP, CRM, spreadsheets and file shares flow into one data foundation that is kept in sync. Nobody maintains a copy by hand.
Sections with source, version, validity and read rights. Old versions stay out, permissions stay intact.
Fill rate, duplicates and contradictions per field, before and after. From then on, fixed rules check every sync.
50 %
Customers live in the ERP, in the CRM and in the sales team's spreadsheet, articles in the supplier catalogue as well. Each source spells them a little differently, none is complete, and every analysis starts with manual work.
Bitkom, Sept. 2026: half of the companies using AI name data preparation a major cost driver (603 companies with 20 or more employees)60 %
The use case was clear, but the fields were empty, duplicated or contradictory. The model was never the problem.
Gartner, Feb. 2025, forecast: through 2026, organisations will abandon 60 % of AI projects unsupported by AI-ready data (248 respondents)Manuals and policies sit in several versions on file shares and in SharePoint. Without version, validity and read rights on every section, a knowledge system answers from the wrong version or shows what the person may not read.
Merge
We read ERP, CRM, spreadsheets and file shares through an interface or an export and match the entries. What belongs together becomes one record; which source leads for which field is your team's call.
Clean
We fill mandatory fields, make spellings and units consistent and set the rules every field has to meet. What was never captured we name, rather than invent it.
Addresses, units, date formats and descriptions in one form. "pcs", "pc." and "pieces" become one unit.
Same VAT ID, similar name, different number: the system proposes, your team decides which record leads.
Mandatory fields, formats and plausibility checks run with every sync. Violations land on a list, not in the ERP.
Checked on every sync
VAT ID present and in a valid formatRule 1
Postcode matches the townRule 2
14 violations since yesterdayto review
For knowledge systems
We split manuals, policies and case files into sections that can answer a question on their own. Each carries source, version, validity and read rights, and a sync keeps it current.
Policies and manuals
Warranty policy v3.docxcurrent
Warranty policy v2.docxold version
Maintenance manual HX-200.pdf84 pages
4. Deadlines
Warranty 24 months from acceptance.Section 1
Version 3, valid from 1 March 2026, read by servicecarried over
Version 2 from 2024replaced, not indexed
Approaches
Data preparation rarely needs a big tool. Which route fits depends on your systems, the volume of data and who maintains it later.
Our stack, run in Germany
Managed, Microsoft
Open source or cloud, sync
Self-hosted, workflow automation
In the system you have
What the system takes on
What your team takes on
Data & documents
Does the data still sit in scans, PDFs, drawings or contracts? Then document intelligence reads it, and this is where the fields go next.
Should questions be answered from the prepared text, with the passage they came from? Then a knowledge system on this foundation.
Should clean records always take the same route between CRM, ERP and approval? Then workflow automation.
From the first call to a system in operation. Each phase stands on its own; none ties you to the next.
free
30 minutes. We work out where AI fits for you and whether an AI strategy is worth it.
€1,990
AI strategy for mid-sized companies: which of your processes suit AI, what the build costs and brings back, and which ones are not worth the effort.
from €5,000
AI pilot at a fixed price: one use case, fixed scope and acceptance criteria in writing up front. Built in your systems, tested with your data.
from 20% a year
Running AI systems: hosting in Germany, monitoring, model and prompt maintenance, ongoing measurement, and the next use case on the same foundation.
Frequent questions
The questions that almost always come up.
Against the use case, not in the abstract. We look at the fields it needs: are they filled, are they consistent, do they exist in every system involved under the same key. What is missing we name, with the effort to fill it.
No, but it is work with a cost. We state what is missing, what exists twice and what contradicts itself before building. An analysis on unclear data is worse than none, because people believe it.
Then Document Intelligence reads it first, field by field and with a confidence score. This is where it continues: the fields read are matched against your master data and taken into the shared data base.
Through a sync that pulls in changes from the sources, hourly, nightly or on every change depending on the source. The quality rules run every time, and replaced or deleted documents are removed from the knowledge system as well.
Not at the start. The entry point is one question from two or three sources, or one area of master data such as customers or articles. A warehouse earns its place when several departments calculate the same figures differently. Then it is the solution, not the entry point.
In your systems, or in an environment that belongs to you. Processing in German data centres or inside your own infrastructure, under a data processing agreement under Article 28 GDPR before any data moves.
The next step
30 minutes, free and without obligation.
