AI consulting
AI consulting:
transform the work before adding more tools
Spentia helps mid-sized companies and large organisations define their AI strategy, identify uses that create real value and prepare deployment. Our starting point is how work actually gets done, the decisions people make and the evidence behind them, rather than a tool or a catalogue of use cases.
Entirely senior team
Vendor-independent advice
Data hosted in France
Value quantified before investment
01 — THE ROLE
What does an AI consulting firm do?
An AI consulting firm helps a business decide why, where and how to use artificial intelligence. It evaluates opportunities and risks, defines strategy and priorities, organises data and governance, and supports the integration of AI into processes and daily work. Its role differs from that of a development agency or IT services firm: it structures and informs decisions before solutions are built.
Strategy and ambition
Connect AI to business objectives.
Maturity and readiness
Establish the starting point using verified evidence.
Use cases and value
Derive priorities from actual work and quantify their full costs.
Data and technology
Check what your IT systems can actually support.
Governance and sovereignty
Set rules for use, compliance and dependencies on suppliers.
Adoption and organisation
Define how people and machines share the work.
02 — OUR POSITION
AI will not transform a business
by decree
Too often, the sequence starts with choosing a tool, launching a pilot and seeking adoption, before defining the return expected.
Three mistakes recur: starting with tools before defining the decisions they will support; selecting use cases from industry catalogues, far from the tasks teams actually perform; and measuring activated licences and trained users instead of value created.
Putting the questions back in order starts with examining what the current sequence hides: the gap between leadership decisions and what happens in daily work.
03 — UNDERSTANDING DAILY WORK
The gap between AI strategy and daily work
often stays hidden
A BCG Henderson Institute and Columbia Business School survey of 1,399 respondents in August 2025 found a marked perception gap.
76%
of leaders believe their teams are enthusiastic about AI
31%
of individual contributors share that view of employee enthusiasm
51 percentage points
gap in agreement that employees are well informed about AI strategy and tools
7×
employee-centric organisations were seven times as likely to be rated AI-mature in the survey; this is an association, not proof of causation
A self-reported assessment can miss this perception gap. A framework completed in a management workshop captures the views of its participants. It may overlook the difference between activated tools and useful applications, as well as unofficial AI use.
Unofficial use deserves to be understood beyond individual fault. Generative AI made first drafts widely accessible before usage rules could catch up. Shadow AI can reveal a gap between available capabilities and the rules governing them. Understanding those practices helps identify both risks and opportunities.
More reporting will not close this gap.
It requires measuring three realities that assessment frameworks often miss.
Three realities shape
every AI transformation
How AI is actually used
Company-provided and personal tools, task by task. Frequency of use, data exposure, time saved and time spent checking or reworking outputs. The difference between an activated licence and productive use.
How work actually gets done
Tasks as teams actually perform them: workloads, friction, exceptions, workarounds and practices that already work. Relevant use cases follow from this evidence. Social and technical systems need to improve together; otherwise, both can deteriorate.
Decision-making capacity
The decision-making process that can support or derail deployment: who decides, which signals are escalated, who resolves issues when the tool is wrong, and who can correct or stop it. This dimension helps explain whether pilots can survive beyond experimentation.
04 — AI CONSULTING
Our approach to AI consulting
rests on five principles
Start with decisions, not tools.
Start with decisions, not tools.
Every existing or proposed use must identify the decision it supports. A tool with no clear decision to support remains a gadget, however widely it is activated.
Invite everyone within scope, rather than a sample.
Invite everyone within scope, rather than a sample.
Our interview system invites everyone within scope to an individual, confidential and anonymous conversation. Comprehensive coverage and anonymity are essential to an accurate picture.
Distinguish what is reported from what is demonstrated.
Distinguish what is reported from what is demonstrated.
We separate claims, documents, records of execution and verified results. No strong conclusion rests on self-reported information alone.
Choose whether to remove, simplify, automate, augment or defer.
Choose whether to remove, simplify, automate, augment or defer.
AI is one possible response to friction. Automating a flawed process reproduces its flaws at scale. Deferring action where value has not yet been demonstrated is a valid conclusion.
Calculate value after all costs.
Calculate value after all costs.
Licences, model usage, integration, human oversight and the cost of errors: value is calculated after these costs and validated with finance.
05 — OUR PRINCIPLES IN PRACTICE
From assessment
to scale
06 — THE AI MATURITY ASSESSMENT
05 — DELIVERABLES
Where to start:
establish your starting point.
Spentia’s AI maturity assessment brings together leadership objectives, teams’ actual use of AI, existing initiatives and available evidence. It identifies barriers, differences in perception, priority use cases and a 12–24-month roadmap. The assessment takes four weeks within the agreed scope.
07 — COMPARING FIRMS
How to choose
your AI consulting firm
Does the firm start with your strategy or its own solutions?
How does it learn what teams actually experience?
How does it verify what people report?
Can it recommend something other than AI?
Is it independent of vendors?
How does it calculate value and full costs?
Who will actually carry out the engagement?
08 — WHEN TO ACT
When should you engage
an AI consulting firm?
Before investing widely in AI tools
To distinguish real needs from trends before committing licence budgets.
When pilots multiply without demonstrated value
To prioritise, stop what should be stopped and prepare to scale what deserves it.
When teams already use AI without a shared framework
To govern actual use without stifling initiative, and build an inventory that supports the review of applicable requirements.
Before deploying agents that can take action
To define permissions, oversight, controls and stop mechanisms before going live.
When leadership cannot agree priorities
To bring ambition, work, data, risks and value into a single decision.
When a transformation changes processes
A merger, reorganisation, change in business model or new IT system is a good time to rethink how work is shared.
09 — FREQUENTLY ASKED QUESTIONS
Frequently asked questions about AI consulting
How do an AI consulting firm, AI agency and IT services firm differ?
How do an AI consulting firm, AI agency and IT services firm differ?
A consulting firm structures and informs decisions. An agency designs and develops a defined solution. An IT services firm integrates and maintains systems at scale. The three often work in sequence within the same programme.
How much does an AI consulting engagement cost?
How much does an AI consulting engagement cost?
The cost depends on scope and depth of analysis. Spentia works for a fixed fee agreed during scoping, with no additional charges for extra consulting days. Documented value must exceed the investment.
How do you identify priority AI use cases?
How do you identify priority AI use cases?
By starting with actual tasks rather than an industry catalogue. Use cases follow from the work described by the people doing it, then are ranked by net value and feasibility.
How do you measure AI return on investment?
How do you measure AI return on investment?
By accounting for all costs: licences, model usage, integration, human oversight and checking, and errors. Value is calculated after these costs, measured against a recorded baseline and validated with finance.
How should unofficial AI use be governed?
How should unofficial AI use be governed?
Start by documenting it. Shadow AI reflects rules that have not kept pace with capabilities, rather than a failure by teams. Anonymous interviews reveal these uses. Governance follows without stifling initiative.
How is return on investment measured?
How is return on investment measured?
In two stages. Before deployment, we record the baseline—time spent, volumes and lead times—so gains can be demonstrated. Each workstream then records quantified results and dates in its tracking sheet, with progress reviewed monthly. We measure the time freed up and how it is used: the share devoted to activities that create value, such as advice, customer relationships and creative work, becomes a tracked indicator.
Does the EU AI Act apply to every business?
Does the EU AI Act apply to every business?
The AI Act may apply when a business provides or deploys AI systems within its scope. Obligations depend on the organisation’s role, the type of system and the risks involved. Implementation dates also vary by requirement. Start by identifying your AI uses and responsibilities, then review the applicable requirements with your legal and data protection specialists.
Do AI agents require specific deployment support?
Do AI agents require specific deployment support?
Yes. An agent can act rather than only suggest. The key questions are which permissions it receives, how people supervise it and how it can be stopped. These arrangements must be defined and tested before going live.
