AI data readiness: what may your AI be built on?
Datengrundbuch keeps your data domains like plots in a land register: with owner, quality, encumbrances and a planning enquiry for every AI project. Before a project starts, you see what data it stands on.
What the tool shows
Spot ownerless data before an AI project builds on it.
Most AI projects fail not because of the model, but because of data nobody feels responsible for. Datengrundbuch shows ownership, quality and legacy burdens of every data domain at a glance.
You enter
- +Data domains per business unit with their leading systems
- +Owner, steward and business value from 1 to 5
- +Quality by completeness, accuracy, timeliness and uniqueness
- +AI projects with the data they need and their minimum quality
You get
- →3D cadastral map: height shows business value, colour shows quality
- →Register extract for each domain with owner, encumbrances and projects
- →Decision for each AI project: approvable, with conditions or not
- →One-page data situation report with effort in person-days
Method
How Datengrundbuch calculates.
Every figure can be traced. The tool gives the formula for every number; here are the most important ones.
Calculation
- +Quality index = 0.3 completeness + 0.3 accuracy + 0.2 timeliness + 0.2 uniqueness
- +Governance score from owner, documentation and legal basis, 0 to 100
- +Lending value = business value × quality; lending ratio across all domains
- +Planning enquiry: no approval without an owner or a legal basis
Assumptions and limits
- →Quality gaps of up to 10 points become a condition, larger ones lead to refusal
- →Heavy encumbrances on a domain become a condition
- →Shortcomings shared by several projects are counted only once
- →A self-assessment, not a data audit in your systems
Example
Example: one of four AI projects is approvable today.
At the fictitious Nordwerk Software GmbH, 2 of 12 data domains are ownerless: price lists and leads, together 18.6% of business value. The support copilot is approvable, the churn forecast only with conditions. The AI quotation assistant fails on the ownerless price lists.
Frequently asked questions
Questions about Datengrundbuch.
What does AI readiness mean for data?
Data is ready for AI when someone is responsible for it, its quality is good enough for the purpose and its use is legally settled. Datengrundbuch checks these three points for each data domain.
What does “ownerless” mean?
A data domain without a registered owner. Nobody decides on definitions, quality and access, so Datengrundbuch does not approve AI projects built on such data.
Do I need access to our systems?
No. You assess ownership, quality and encumbrances yourself; the tool calculates in your browser and transmits nothing. Measuring in your systems is the next step in a consulting engagement.
How does this relate to the EU AI Act?
For high-risk systems the AI Act requires, among other things, data quality and data governance. Datengrundbuch shows where these foundations are missing today. KI-Depot classifies your projects.
What is data ownership?
Clear responsibility for a data domain: a person from the business decides on definitions, quality targets and access, a steward implements them. Without that responsibility, data quality remains a project without a sponsor.
How do I measure data quality?
Along four dimensions: completeness, accuracy, timeliness and uniqueness. Datengrundbuch weights them into a quality index from 0 to 100 and compares it with the minimum quality an AI project needs.
From land register to data strategy.
In a free 30-minute initial call we review your assumptions and discuss how the self-assessment becomes a reliable diagnosis.