What the model measures.
How your organisation plans, attracts, develops and retains the people needed for AI work.
Organisational talent maturity
The assessment examines how strategically and consistently an organisation manages AI-related talent. It covers people doing technical, product, governance, learning and leadership work. A high level requires talent practices to support business goals.
The four stages are Reactive, Standardised, Strategic and Optimised. This is an original assessment design based on those stage definitions. CIPD talent-management and workforce-planning guidance informs its scope; CIPD has not designed or validated this assessment.
Six areas, one transparent rule
Strategy and workforce planning; attraction and recruitment; development and internal mobility; reward, engagement and retention; leadership and succession; talent data and improvement.
Select the highest stage that consistently describes each area. The overall reported stage is the lowest of the six selected stages. This conservative rule avoids a strong recruitment process masking a missing retention or succession practice. It is an editorial rule, not a scientifically validated cutoff.
If any answer is “Not enough evidence”, no overall stage is assigned. Unknown is not treated as Reactive. The result still shows the six-area profile and suggested actions. Priorities put evidence gaps first, then the lowest reported areas; ties follow questionnaire order.
The assessment has no peer sample, percentile ranking, predictive validity or certification. Use it for a structured discussion and repeat it against agreed evidence. It should not be used to assess individual employees.
What counts as evidence?
Use current plans, hiring criteria, learning assignments, internal moves, reward decisions, continuity arrangements and reviews that changed a decision. Answer for a defined organisation or business unit, using the last 12 months. A written policy alone does not establish that a practice happens consistently.
How the supporting tools fit
Role guides help define work and development expectations. Their four capability levels describe individual scope of responsibility; they are separate from the organisation’s four talent-management stages.
Salary benchmarks support compensation and retention decisions. Source populations, currencies, dates, statistical measures and role matches are stated per observation. No salary is calculated from a maturity stage.
Market context describes national enterprise AI use and the broader ICT workforce. It does not measure an organisation’s talent practices or an individual’s capability.
Salary sources, annualisation and estimates
Salary evidence covers seven countries. Sweden’s monthly range is multiplied by 12; this is an annual equivalent, not total compensation. Currency conversion is not applied. UK advert percentiles, Swedish member percentiles, technology-category means and recruiter-guide ranges remain separately labelled.
Irish guide ranges are source estimates; its experience bands are not maturity stages. The Irish table does not explicitly define gross/net or bonus treatment, so pay inclusions are marked unspecified. France explicitly reports gross starting fixed pay excluding bonuses and benefits. The data includes sample sizes only where the source provides them.
We do not infer salary premiums for AI, convert contractor day rates to employee salaries or extrapolate salaries into countries without evidence. Sources can be dated, unrepresentative or affected by the mix of advertised roles. Treat the figures as planning inputs and check current local offers.
Country readiness proxy: formula and coverage
The market view covers 41 markets: all 27 EU members and 14 non-EU markets. Only 31 have both 2025 inputs. Country scope and missing observations are listed in the market table. Data was retrieved on 14 September 2026.
Adoption component = min(AI adoption / 50%, 1) × 100. Technical component = min(ICT employment share / 10%, 1) × 100. The default proxy averages the two equally. The 50% and 10% anchors and weights are editorial choices. A score of 80 is not “80% mature”.
The table offers 30/70 and 70/30 weights as sensitivity checks. Missing inputs produce no score. Unrounded values determine sorting. Source estimates and other status flags are retained. The proxy is not validated against hiring outcomes and is entirely separate from the organisational assessment.
Country definitions, source flags and scenarios
Enterprise adoption covers selected activities and employers with at least ten people. It includes purchased tools. ICT specialists include many occupations outside AI. Technology-use categories overlap; do not sum them. In-house development is a share of AI users, while the expertise barrier concerns enterprises that considered AI without using it.
“e” denotes a source estimate, “b” a break in series, “p” provisional data, “u” low reliability and “d” a different definition. Prior-year figures are kept separate. Changes between 2024 and 2025 are descriptive and may reflect changed wording or survey coverage.
Country pages retain mechanical 2026 adoption scenarios: 2025 adoption plus 0.5, 1 or 1.5 times the 2024–2025 percentage-point change, capped at 0–100%. These are labelled planning scenarios, not official forecasts, confidence intervals or inputs to the proxy.
Sources behind the model
Privacy and corrections
The assessment runs in your browser without sending or storing your answers. Reloading clears them. There is no analytics or sign-up form. Ordinary hosting access logs may still be processed. External websites have their own policies.
Email johannes@sundlo.com with corrections or source suggestions. This is a dated snapshot with no automated update schedule.