AI data readiness check
Is your data ready for AI?
Anyone wanting to use AI on their own data – an assistant that knows the company knowledge, an AI search across documents – rarely fails because of the model and almost always because of the data foundation. In 8 short questions, this check shows you where you stand and what your biggest lever is. Honestly assessed, no sign-up.
The AI data readiness check assesses how well a company's data foundation is suited to running AI on its own data (such as RAG, AI search, or an AI assistant on company knowledge) – across five dimensions: findability, quality, format, access and data protection.
Your data foundation
Your data readiness
Overall readiness
64/100
Partly ready
A foundation is there, but there are clear building sites. An AI project is possible – best focused on the area where your data is already strongest, while the weakest dimension is cleaned up in parallel. The breakdown below shows where to start first.
The five dimensions in detail
Your biggest lever: Quality & currency
Your biggest lever is data quality. On outdated or contradictory data, AI confidently gives wrong answers – garbage in, garbage out. Before using AI, cleaning up pays off: remove duplicates, retire outdated content, unify terminology.
Your strength: Findability & structure
Is the knowledge in one place and ordered – or scattered and chaotic?
Do you really want to use AI on your own data?
How this check evaluates
- Five dimensions: findability & structure, quality & currency, format & machine-readability, access & interfaces, and data protection context – derived from the factors that make AI on your own data fail in practice (see our article on RAG instead of fine-tuning).
- Each answer scores points for exactly one dimension; the overall value is the equally weighted average of the five dimensions (0–100). The evaluation runs deterministically in your browser, without AI and without any inputs being sent.
- The data protection context doesn't assess data quality but sensitivity: a low value means “needs a compliant architecture”, not “bad data”.
- Deliberately honest: instead of a flattering overall figure, the weakest dimension is named openly as the biggest lever – because that's exactly where it's decided whether an AI project holds up.
This check provides a self-assessment for orientation, not a data audit. Actual suitability depends on details we clarify in a conversation or a closer analysis.
Frequently asked questions about the data readiness check
What is this check for – and what not?
It's for the case where you want to use AI on your own data: an assistant that knows the company knowledge, an AI search across documents, automated answers based on internal content. It assesses your data foundation for that. It's not a general AI maturity check for your whole company – for that there's the more detailed Readiness Assessment.
How does it differ from the Readiness Assessment?
The Readiness Assessment looks at your company broadly: processes, potential, team, implementation readiness. This check deliberately goes deep on a single question – is your data foundation suited to AI on your own data? If you already know you want to go in that direction, this check is the more precise one.
Why is the data foundation so decisive?
Because AI projects on your own data rarely fail because of the model and almost always because of the data: it's scattered, outdated, contradictory, trapped in paper or not cleanly accessible legally. AI can only answer as well as its knowledge base. That's why an honest look at the data pays off before money goes into a model.
What does a low value in the data protection context mean?
Not that your data is bad – only that it's sensitive. Personal or highly sensitive data doesn't rule out AI, but it requires a compliant architecture, often hosted locally or in the EU rather than in an arbitrary cloud. That's very feasible and belongs in the planning from the start. More on this in our article on GDPR and AI.
What do I do with the result?
Use it as orientation on where to start first. If the check shows a good basis, the next step is to define the concrete use case and build a prototype – Custom Applications supports that. If it shows gaps, you know which data building site comes first, instead of putting money into an AI project that would fail because of it.