A nine-dimension framework — refined over years at Volkswagen Commercial Vehicles, updated for the GenAI era, and now a free 25-question assessment.
When I was responsible for data and AI strategy at Volkswagen Commercial Vehicles, we ran the same exercise every two years: an honest, structured measurement of where the entire organisation actually stood on data readiness. Not a survey, not a self-assessment workshop — a proper baseline across nine dimensions, benchmarked against previous years and against comparable manufacturers.
That discipline is what separated the initiatives that scaled from the ones that quietly died in a PowerPoint. Over the past year we have taken that same framework, updated it for the GenAI era, and turned it into a free 25-question online assessment. This article explains why the framework works, what the nine dimensions actually measure, and what you can do with the results.
Without a baseline measurement, AI investment turns into wishful thinking.
Most organisations do not fail at GenAI because the technology is too complex. They fail because nobody has honestly measured whether the foundation — data, skills, governance, organisation — is actually ready to carry the weight of production AI. A clear assessment across nine dimensions does not just show the status quo. It forces the difficult prioritisation conversation: where is the single biggest lever, and who owns fixing it before the next investment cycle?
Why most GenAI initiatives skip the baseline
At Volkswagen Commercial Vehicles I saw a pattern I have since seen at BOMAG, at a large mobility-sector financial services provider, and at half a dozen mid-caps: leadership announces an AI strategy, budget is committed, pilots are launched — and eighteen months later the retrospective is uncomfortable because the pilots did not translate into value.
The reason is almost never the model. It is that pilots were selected before anyone knew whether the underlying data was usable, whether the platform team had capacity, whether governance would clear the model into production, or whether the organisation had the change appetite to actually adopt what was built.
Measurement fixes that. Not because a radar chart magically makes problems disappear, but because it moves the conversation from “we should really do more AI” to “our biggest gap is Dimension 3 — let us close that before we spend on Dimension 8.”
The nine dimensions
The framework we refined at ultra mAinds covers nine dimensions. They are deliberately broad, because GenAI-readiness is a systems problem, not a technology problem.
- Data. How accessible, clean, and well-catalogued is your operational data? Can a new use case find and consume the right sources without a six-month integration project?
- Infrastructure. Do you have the compute, storage, and networking to run modern AI workloads securely — including private connectivity to hosted foundation models where required?
- Skills. Which roles exist in-house — data engineer, ML engineer, prompt engineer, AI product owner — and how quickly can you hire or train for the gaps?
- Governance. Are model-approval, data-usage, risk-management, and audit processes explicit, or do they exist as tribal knowledge?
- Use cases. Is there a live, prioritised backlog with ROI estimates, or a list of ideas without owners?
- Organisation. Where does AI live — a central platform team, embedded in every business unit, or a mix — and are the incentives aligned with the operating model?
- Tooling. Which platforms, MLOps stacks, and vendor relationships are in place, and where are the licence, lock-in, and total-cost-of-ownership traps?
- Ethics & compliance. How are the EU AI Act, GDPR, sector regulation, and internal ethics guidelines translated into operational checkpoints — not just policy documents?
- Vision. Is there a clear, communicated three-year picture of what AI-driven value creation looks like, and does the leadership team tell the same story?
Every dimension is scored on a four-point maturity scale — from exploring to leading — with concrete evidence expected at each level. The scoring matters less than the internal conversation it forces.
What happens when you actually do it
Three things surface reliably every time we run this assessment with a new organisation.
One dimension is significantly weaker than everyone believed. Usually data or governance. The room goes quiet for a moment, and then the CTO says something like “we have been talking about that for two years.”
One dimension is significantly stronger than the leadership team gave itself credit for. Usually skills or tooling, because individual excellence in the trenches has quietly outpaced the strategic picture at the top.
The gap between dimensions is often larger than the average. Averaging “we are a 2.4 across the board” hides that you are a 3.5 on infrastructure and a 1.2 on data. The average does not tell you where to invest. The gap does.
At Volkswagen Commercial Vehicles the biannual cadence made the delta between assessments the actual management artefact. Not the absolute score. The direction of travel.
When the framework does not fit
Full transparency: the nine-dimension framework is not the right tool if you are still deciding whether AI is strategically relevant to your business at all. That is a different, earlier conversation, and no assessment can answer it.
It is also not the right tool for measuring a single use case in production — that needs its own KPIs (accuracy, latency, adoption, business value delivered). The framework measures organisational readiness to run and scale AI, not the performance of a specific model in production.
Where it works reliably is the space between “we have an AI strategy on paper” and “AI is a normal part of how we build software and make decisions.” That is where most enterprises sit today, and where the biggest value from measurement is unlocked.
Take the assessment
We took the nine-dimension framework, consolidated it into 25 questions across the core categories, and made it free.
What you get:
- 25 questions, roughly 10 minutes to complete honestly
- An immediate radar chart of your scores across the categories
- A personalised PDF report by email — concrete recommendations plus benchmark comparisons against tech mid-caps and large enterprises
- No sales call unless you ask for one
If you already have a strategy team, run the assessment with three or four people independently and compare answers. The disagreements are the most valuable output.
What comes next
This is the first article in a ten-part series on how agentic AI is actually changing enterprise software development — with real customer stories from BOMAG, from a mobility-sector financial services provider, from a Japanese automotive corporation, from a German SaaS provider, and from our own agent-orchestrator running twenty-four hours a day on a real ticket board.
Every article ships with a concrete asset — a template, a diagram, or a public demo repository. No generic “what is RAG” content. Only what we have actually built, measured, and learned.
