Published
Updated
AI assessment and AI maturity
AI assessments help an organisation decide where it stands before committing resources. Common dimensions include strategy, data, technology, skills, governance and culture. A maturity score can organise that discussion, but needs supporting evidence and a clear connection to investment decisions.
What a general score cannot tell you
Two organisations with similar readiness scores may need different solutions. One could benefit from automating incoming-request qualification. Another may need to preserve that conversation while improving technical-file preparation.
Readiness and usefulness are different questions. Identify tasks that are repetitive, demanding or valuable to customers, and understand what would change if AI took part. A general score cannot substitute for that investigation.
From pilots to sustained use
Deloitte’s 2026 global research reported that only 25% of respondents had moved at least 40% of their AI pilots into production. That finding illustrates the distinction between experimentation and routine operation; it is not a universal failure rate.
Also investigate unofficial use. Employees may already have found useful applications outside company-provided tools. These practices can reveal both unmet needs and risks involving data, verification or personal accounts. Make it possible to describe them candidly before deciding how to govern them.
A method that starts with tasks
1. Define the decisions. Which areas deserve investment? Which activities should remain human-led? Which pilots should stop? These questions determine scope.
2. Inventory current use. Include pilots, licences and unofficial tools. Explain confidentiality and how the information will be used.
3. Map actual tasks. Ask people in each role what takes time, what repeats, what creates difficulty and what they see as their most valuable contribution. Invite everyone in scope and record gaps in participation.
4. Check the prerequisites for each case. Assess the relevant data, integration, skills and governance. ‘Do we have the information needed for this task?’ is more useful than a general verdict on data readiness.
5. Build a decision matrix. Consider automation, assistance, process redesign or protection of the human contribution. For each option, record prerequisites, expected value, risks and a proposed owner.
Organisation and assessments
Make organisational decisions with a clear assessment
Our assessments identify practical changes to help your teams and processes work together.
Useful deliverables
Expect a prioritised portfolio of use cases grounded in tasks, a view of prerequisites for each case, an inventory of current uses and risks, and a phased plan with decision points. The inventory supports governance and assessment of applicable legal duties; it does not establish compliance by itself.
A maturity chart can remain a useful summary. It should be accompanied by the evidence and choices behind the score.
Readiness, assessed in context
Data quality requirements depend on the intended result and the consequences of error. A drafting assistant and a forecasting system need different checks. Neither should be used beyond what its data and review arrangements can support.
Training should address the changed tasks, including verification and when to take control. Governance needs clear owners and timely decisions close to the workflow. Technical integration and maintenance need explicit resources. Adoption improves when teams can see a useful change and understand their responsibilities.
How to judge the result
Show the portfolio to the people doing the work. They should recognise the tasks and constraints. Leaders should be able to identify what to do, for whom, in what order and under which conditions. The assessment is useful when both groups can follow the evidence behind those choices.
Further reading
Frequently asked questions
How does an AI assessment differ from a maturity assessment?
How does an AI assessment differ from a maturity assessment?
A maturity assessment examines readiness. A broader AI assessment also identifies which tasks to change and what value to expect. Good work connects the two and produces decisions supported by evidence.
How long does an AI assessment take?
How long does an AI assessment take?
Timing depends on scope and access to evidence. Allow several weeks for task mapping, technical checks and decisions. Spentia scopes its assessment around a four-week programme; participation and required evidence are agreed at the outset.
Do we need perfect data before using AI?
Do we need perfect data before using AI?
You need data that is fit for the specific use and its risks. Assess completeness, reliability and access for each case, along with review arrangements. A company-wide data-cleaning programme is not a prerequisite for every possible application.
Should IT or business teams lead the assessment?
Should IT or business teams lead the assessment?
Both should contribute, with direct input from people doing the tasks. IT assesses systems, integration and technical controls; business teams describe the work and intended outcomes. Clear shared ownership connects the two.
Sources
Deloitte, The State of AI in the Enterprise: The Untapped Edge, 2026. Report overview.
