As organisations move artificial intelligence beyond isolated experiments, judging whether they are ready to use it is becoming less about technical capability alone and more about whether management, governance, data, and operating processes can support repeatable deployment.

AI Sweden has updated the AI maturity-assessment process it has offered partner organisations since 2021, giving adoption and responsible AI a larger role alongside technical capability. The original method was developed by Germany’s appliedAI, with AI Sweden contributing to the revised version.

The process starts with a questionnaire designed to establish how an organisation governs and uses AI. Management then receives analysis and recommended activities, measures, and decisions, followed by a workshop that brings employees from different functions together to examine the results and agree practical priorities.

The revised emphasis reflects how enterprise AI has developed since 2021. Capable models, cloud infrastructure, and developer tools can now be acquired quickly, while deciding where AI should be used, which data it can access, who owns a deployment, and whether the resulting workflow is producing useful outcomes remains slower organisational work.

Adoption enters the maturity score

Mälarenergi, the Swedish energy and infrastructure company, became the first organisation to use the updated process. One of its main conclusions was that the company needed an organisation-wide AI ambition so that individual projects could be assessed against a clearer strategic direction.

That problem becomes more visible as generative AI arrives from several directions at once. Central technology teams may build approved applications while staff use general-purpose assistants, software suppliers add AI functions to existing products, and individual departments buy specialist tools. An organisation can therefore accumulate AI capability without establishing consistent rules for how the technology enters day-to-day work.

Östgötatrafiken, Region Östergötland’s public transport company, completed the assessment for the third time in spring 2026. Repeated use turns the exercise into something closer to a management cycle, because governance changes, new deployments, and workforce experience can be measured against an earlier baseline.

The workshop element also limits one familiar weakness in maturity frameworks. Long capability inventories can easily produce equally long improvement programmes, whereas the revised process encourages organisations to identify a smaller number of focus areas before the next assessment.

Governance becomes part of production

The stronger emphasis on responsible AI arrives as European organisations contend with data protection, cyber risk, documentation, human oversight, and obligations created by the EU AI Act alongside the commercial questions of cost and performance.

Those controls are difficult to add after deployment because they influence system design. An organisation that needs to document model behaviour, restrict access to confidential information, retain human review, or monitor outputs has to build those requirements into workflows and architecture rather than maintain governance as a separate policy exercise.

Maturity scores still have an obvious limit: they do not demonstrate that AI investment is generating economic value. A business can establish committees, standardise infrastructure, train employees, and document models while continuing to run systems that save little time or fail to improve the underlying process.

The useful evidence therefore sits between organisational readiness and operating results. Governance determines whether systems can be deployed safely and repeatedly, while adoption measures need to show whether employees use them, whether processes change, and whether any productivity or service improvement survives beyond the pilot stage.

AI Sweden’s revised method does not remove that measurement problem, although it reflects a market in which access to models is becoming less distinctive than the ability to deploy them coherently. As technical scarcity falls, organisational maturity is increasingly determined by whether companies can turn available AI into governed working systems without confusing the presence of new tools with proof that those tools are improving the organisation.