Agentic AI maturity · Governance, strategy, culture, data and IT readiness
Your capacity to let AI act, measured theme by theme and turned into a roadmap.
10 themes, a 5-level scale. And the action that moves each level to the next.
The framework’s 10 themes, already written from L1 to L5. One company, one business unit, or 300 at once.
Agentic AI maturity · Governance, strategy, culture, data and IT readiness
10 themes, 5-level scale.
Nordhavn Industries
53 / 100
They measure their maturity with Datamensio
An example
This could be your situation.
Take one company as an example: three sites, three spreadsheets, no shared answer.
Nobody can consolidate.
Nordhavn Industries, 2,400 people in Hamburg, Lyon and Porto. A client asks where the group stands. Each site answers in its own spreadsheet, with its own scales.
Three weeks, a single base.
One Agentic AI maturity framework (Datamensio framework, 5-level CMMI scale) assessment launched across all three sites at once, from the managers’ interview notes. The framework was already written, its 10 themes and levels L1 to L5 too.
Two costs avoided before being committed.
A score of 53 out of 100, with the gap concentrated on three themes. The AI companion spotted that two actions duplicated those of another audit. The committee report took one sentence to request.
What it saved them
- 3sites measured on the same base, instead of three questionnaires to reconcile
- 2duplicate actions caught before the spend
- 1committee report, with no manual rework
These figures are an example. They could be yours.
The standard imposes processes. Datamensio says where you stand.
01
The framework is already written
Themes, questions and levels L1 to L5, all written. You do not start from an empty spreadsheet.
02
The score lands the same day
Online, by self-assessment link or in interview. Theme by theme, comparable over time.
03
The gap becomes a costed plan
Every step up carries its action. The AI prioritises on expected effect, not on the order of the standard.
04
Progress can be demonstrated
Campaign after campaign, against your target and against your own past. That is what your board asks for.
The maturity scale
One level, the next, and the action that links the two.
This mechanism, a level, a level above, and the action connecting the two, is what turns an observation into a trajectory.
Is the action scope of agents within business systems defined and controlled?
- N1
No scope defined. Agents use the access of the person or team who set them up, with no written boundaries.
- N2
Some rules exist, owned by the team that built the agent. They are not written the same way from one project to another.
- N3
Each agent has a documented scope, a dedicated technical identity and a designated owner. Deviations are occasional.
- N4
Scopes are technically enforced, actions are logged and traceable to the agent, with periodic review of permissions.
- N5
Scopes evolve based on incidents and new uses, with documented tracking of revisions and comparison across entities.
Action to move from L2 to L3
Create a standardised agent record (authorised tasks, accessible systems, dedicated technical identity, business owner, stop conditions), apply it to all agents in service and validate the records at the monthly architecture committee.
« With Datamensio, we meet our objectives far more efficiently. The ERDF inspection services and our supervising ministry particularly appreciated an approach that gives them reliable data. »
Director, CCI 94CCI Île-de-France
« We believe this is the most suitable solution to scale our transformation project and measure impact according to our needs. »
Maja SucekChief Operating Officer, Interreg Danube
Rarely on its own
Frameworks combine. Put several together to cover your business, or have the AI write yours.
Take your first measurement
What this framework covers
Agentic AI differs from content-generating AI on one simple point: the agent does not suggest, it acts. It chains steps, calls tools, writes to systems, triggers tasks and reports back. This requires machine identities, delimited action scopes, usable logging, human checkpoints and reliable data at the moment of the call. The framework assesses these conditions separately, because an organisation can be advanced on experimentation and well behind on execution.
In practice, the subject is hard to govern because it cuts across IT, business units, data and legal. Who authorises an agent to write to a production system, and within what scope? What happens when an agent chains ten actions and the third one was wrong: is there a trace, a stopping point, a designated owner? Do business teams know which decisions they are delegating and which they are keeping? Without answers, projects remain stuck at the demo stage, not for lack of technology but for lack of an execution framework.
A common confusion is worth clearing up: agentic AI is not a more powerful model, it is an architecture. Performance comes from orchestration, access to tools, the quality of the data used and the design of control points. Another point of context: organisations that have structured their data governance and access management move noticeably faster on agents, because the foundation is already in place. Conversely, a poorly documented data estate turns every agent into an integration project.
The framework does not measure compliance, it locates a transformation trajectory. It does not end in a binary verdict but in a maturity level and the action that moves each theme to the next level. This reading allows a choice to be made: consolidate the data foundation before industrialising, or first secure the action scopes of agents already in service. The gap between the observed score and the target set by management generates the action plan, which the AI groups into a prioritised roadmap.
The framework is ready to use and belongs to you. You adjust the themes, questions and levels according to your sector and organisation. The AI refines the wording, aligns levels with the CMMI method, or builds a variant from your own documents: AI usage policy, application map, IT master plan. Assessments can then be run business unit by business unit, then compared with each other and with previous rounds.
Reference standard: Agentic AI maturity framework (Datamensio framework, 5-level CMMI scale)
The themes assessed
Strategy and use cases
Existence of an explicit trajectory, selection and prioritisation of agentic use cases, arbitration criteria, alignment with the strategic plan, measurement of expected value.
Governance and responsibilities
Decision-making body, designated roles for authorising an agent, definition of action scopes, escalation rules, review of agents in service.
Culture and skills
Level of business awareness, training of technical teams in orchestration, ability to formulate and challenge an agent’s decision, change management.
Data estate and quality
Source mapping, semantic documentation, freshness and reliability at the moment of the call, management of data access rights, traceability of usage.
IT readiness
Exposure of business functions through interfaces, representative test environments, management of machine identities, usable logging, capacity for rollback and cancellation.
Agent design and orchestration
Delimitation of tasks assigned, choice and selection of models, management of tools called, human checkpoints, handling of unforeseen cases.
Evaluation and monitoring in operation
Test sets, action quality indicators, drift detection, incident review, correction loop.
Security and risk control
Agent permissions, scope segregation, protection against prompt manipulation, handling of sensitive data, fallback plan.
Responsible use and legal framework
Internal usage policy, information of affected individuals, system documentation, alignment with applicable obligations and contractual commitments.
Industrialisation and scaling
Component reuse, production release cycle, tracking of execution costs, common foundation across business units, cross-project capitalisation.
A short version of the framework is available for the online self-assessment.
Frequently asked questions
How does this assessment differ from an audit?
An audit checks whether requirements are met and concludes with a gap. This framework is not certifying: it locates your practices on five levels and indicates the action that advances each theme. The output is not a verdict, it is a costed roadmap.
Do we need to have already deployed agents to take part?
No. An organisation at the exploration stage gets a low score on the operational themes and a useful baseline on strategy, data and IT. This is often when the assessment has the most value, before budgets are committed.
How does this differ from a general AI maturity assessment?
General AI maturity covers all uses, including analysis and content generation. This framework focuses on autonomy of action: access to tools, scopes, traceability, human rollback, IT readiness. The two complement each other and can be consolidated into a cross-cutting roadmap.
How long does the assessment take?
The short version takes around thirty minutes to complete with a single respondent. In collaborative mode, involving IT, data, security and business teams, allow one to two weeks, most of the time being spent gathering input from teams.
Can the framework be adapted to our context?
Yes. Themes, questions and levels can be modified, and the AI refines the wording or builds a variant from your internal documents. You can also start from a blank template. The framework remains yours to use.
How can several business units be compared?
Assessments run on the same framework can be compared across entities and with previous rounds. You see where the gaps lie and which actions can be shared. Several assessments can be consolidated into a single roadmap.
Is the action plan costed?
Yes. Each gap between observed score and target opens one or more actions, to which the service catalogue assigns a cost, a timeframe and an estimated impact on the score. You arbitrate on comparable elements.
Where is the data hosted?
In France, with OVH, backed up with Scaleway. No transfer outside the European Union. The AI models used can be selected, including from European solutions.