Skip to content
Technology transformation expertise firm · ARDIANE Group Innovate, secure, succeed

Insights

Insights: our analysis of AI, data and the cloud

What we learn in the field, written for those who decide.

In brief

Management Square's insights bring together the firm's analysis of artificial intelligence, data, cloud and platforms. They are written for the decision-makers who lead these projects in their organisation.

Should you try the BMAD method to organise AI-assisted development?

The BMAD method is a third-party open project, published on GitHub under the MIT licence, that organises AI-assisted development around role-specialised agents producing documents that the next ones take over. Management Square presents its content, what it requires, the cases in which to try it and a way of judging a trial, based on the project's repository and documentation.

Read the analysis
Building a data migration plan and rehearsing it before cutover

A data migration plan organises the move of data from a source system to a target system: what migrates, under which rules, in which window and with which checks. Management Square sets out here a method in six successive steps, from inventory to post-cutover checks and rollback, each closed by a deliverable that the next one uses.

Read the analysis
Cyber resilience best practices, judged on what a test has proven

For Management Square, cyber resilience best practices are those that a test has proven: backups actually restored, business continuity and disaster recovery plans exercised, dependencies known down to suppliers, crisis decisions prepared. The article proposes an assessment in calm times: the signs that separate proven resilience from declared resilience, and what each gap calls for.

Read the analysis
Generative AI observability adds cost, answer quality and drift to latency

Generative AI observability refers to the measures taken on an application based on a language model in production: latency, cost per request, evaluated answer quality, tool call failure rate, drift. Management Square sets out here how to establish each measure, the reader it serves and the blind spot it leaves.

Read the analysis
Setting up agentic AI governance on the basis of an existing AI governance

Agentic AI governance brings together the rules and roles through which an organisation knows which AI agents act in its applications, with what rights, what degree of autonomy and under the responsibility of which person. Management Square proposes here a method for adding it to an existing AI governance: inventory, autonomy levels, designated owner, traces, periodic review, shutdown.

Read the analysis
Data quality management decides where the dashboard only observes

Data quality management refers, by working convention, to the forum and rules through which an organisation chooses the data it monitors, assigns each gap to a person who decides and determines where to correct it. Management Square offers here a guide to the decisions this management must make, and clarifies the gap that separates it from a dashboard.

Read the analysis
The ethical challenges of generative AI read in the public reference texts

An ethical challenge of generative AI is, by convention, a question that the law alone cannot settle: bias, misleading content, copyright, transparency, work, environmental footprint. Management Square reads what the OECD Recommendation on AI, the 2019 European guidelines and Union law say about them, then proposes a way to derive reasoned choices for each use.

Read the analysis
Launching a sustainable IT approach in business, step by step

A sustainable IT approach in business first measures the environmental footprint of its equipment and services, then acts on their lifespan, their design, their hosting and their procurement before reporting on it. Management Square proposes this order and indicates what each step delivers, drawing on the French REEN law and the RGESN.

Read the analysis
What and whom do we depend on? Enterprise digital sovereignty before the executive committee

A company's digital sovereignty is an expression of common usage: Management Square proposes to understand it as the ability to decide on one's digital dependencies (infrastructure, software, data, skills, suppliers). The article follows an executive committee's questions: what and whom the business depends on, what would remain if a dependency ceased, what to protect first, and at what cost.

Read the analysis
GDPR and artificial intelligence in the same project, the articles to read in the text

GDPR and artificial intelligence intersect as soon as an AI project involves the processing of personal data, in training as in use. Management Square reads here the articles of Regulation (EU) 2016/679 that such a project encounters (5, 6, 13 to 15, 22, 25 and 35), what they say, what they leave to be assessed, and how they fit with the AI Regulation.

Read the analysis
Digital operational resilience, as defined by Article 3 of the DORA regulation

Digital operational resilience is, according to Article 3 of Regulation (EU) 2022/2554 (DORA), the ability of a financial entity to build, assure and review its operational integrity and reliability, including throughout disruptions. Management Square reads this definition and the regulation's five pillars here, separating what the text sets from what it leaves to technical standards.

Read the analysis
A cloud disaster recovery plan is built application by application, on the share of recovery that falls to the customer

A cloud disaster recovery plan is the recovery plan for applications hosted by a cloud provider: it describes how to bring them back into service within the time and with the data loss the business accepts. Management Square proposes building it in this order: objectives per application, sharing of responsibilities with the provider, scenarios, backups, failover mode and testing.

Read the analysis
Comparing sovereign cloud or hyperscaler, starting from the data to be hosted

Choosing between a sovereign cloud and a hyperscaler means comparing two families of offerings: those that highlight guarantees on location, applicable law and control of operations, and those of cloud providers on a global scale. Management Square compares them on the law applicable to data, qualification, service catalogue and reversibility, then describes mixed arrangements.

Read the analysis
Carrying out data deduplication without merging in error

Data deduplication consists of identifying the records that refer to the same entity, deciding which to merge and according to which rule, then preventing duplicates from forming again. Management Square proposes here an order for carrying it out: scope and matching rule, normalisation, detection, review of doubtful cases, merging and prevention at data entry.

Read the analysis
Cloud or on-premise: the criteria that decide each application

The cloud or on-premise choice sets resources rented from a provider and consumed on demand against an infrastructure the organisation operates itself; it is decided application by application. Management Square presents here the criteria that decide between public cloud, on-site hosting and the intermediate forms, and how to prepare the exit.

Read the analysis
The dimensions of data quality, from one public framework to another

A data quality dimension is an angle from which data is judged, for example its accuracy or completeness: it is a convention, and the lists vary from one framework to another depending on the object being judged. Management Square compares here three public texts read at source and proposes a way of choosing a list for your organisation.

Read the analysis
Which data quality metrics to track, and how to calculate them

Data quality metrics are calculated measures, with a constant definition, on records or on the handling of their anomalies: rate of conformity to a rule, fill rate, duplicate rate, availability lead time, age of open anomalies. Management Square specifies here, for each one, what it measures, its formula in words, its reader and what it does not say.

Read the analysis
A data governance policy is judged by the arbitrations it writes down

A data governance policy is the text that sets the lasting rules on data: scope, roles, classification, access, arbitration of definitions, handling of gaps. Management Square suggests leaving operating procedures to separate documents and having it approved by executive management or the executive committee; even once approved, it does not on its own establish compliance with a regulation.

Read the analysis
Putting data maturity to the test of four real requests

Data maturity refers here to an organisation's ability to provide, for a decision or a service, data that someone answers for, whose quality is known and that can be obtained within a predictable time. Management Square proposes to place it through four tests (responsibilities, quality, platform, uses), without adopting a published model, and indicates the next step each gap calls for.

Read the analysis
Data owner and data steward, two roles separated by the power to decide

In common usage, the data owner is the business manager who decides for a data domain, and the data steward the person who prepares these decisions and applies them day to day. Management Square compares the two roles here (decision, execution, scope, time allocated) and describes how they fit with the technical custodian, on the IT side.

Read the analysis
Under what conditions is a real-time data architecture worth its cost?

A real-time data architecture processes data as it arrives in order to meet a delay, the one separating an event from the decision it informs, when batch processing can no longer meet it. Management Square reviews the questions of an investment committee: which decision, which delay, which step between batch and stream, which operating burden, which place for batch.

Read the analysis
Settling the "data warehouse or data lake" question on the basis of the uses to serve

The "data warehouse or data lake" question sets the data warehouse, where structure is imposed before loading for known questions, against the data lake, where files are deposited raw and structured on reading. Management Square compares the two on structure, uses, governance and cost, then situates the lakehouse and the coexistence of the two systems.

Read the analysis
The choice between ETL and ELT depends first on where the computing power lies

ETL or ELT: both chain extraction, transformation and loading, and differ in where the transformation runs, before the target or within the target. Management Square proposes deciding flow by flow, on four criteria: where the compute lies, the sensitivity of the data, the team's skills and the pay-as-you-go cost.

Read the analysis
How Article 6 of the European AI regulation identifies a high-risk AI system

A high-risk AI system is, under Article 6 of Regulation (EU) 2024/1689, a system linked to an Annex I product subject to third-party conformity assessment, or a system referred to in Annex III, unless the paragraph 3 derogation applies. Management Square reads here this article and its annexes in the consolidated version of 27 July 2026, and points out what the text leaves to be qualified.

Read the analysis
What a corporate AI charter must settle, what it leaves to other documents and who signs it

A corporate AI charter is the document that tells each user which AI tools they may use, which data they must not enter into them, what they must review and whom to report a deviation to. Management Square sets out what this charter must settle, what belongs in other documents (system register, security policy, contracts) and who signs it, without treating it as a sufficient means of compliance.

Read the analysis
Permissions, tools, content read: AI agent security as a CISO examines it

AI agent security consists of limiting what an agent can reach, whatever the model decides: permissions reduced to the task, a closed list of tools, justified reads, content read treated as untrusted, a trace of every call and a tested shutdown. Management Square reviews the questions a security officer asks before authorising an agent, based on named public sources.

Read the analysis
Who decides the next step? AI agent orchestration, from fixed code to handing over

AI agent orchestration refers to the way an application decides which agents run, in what order and who chooses the next step: the code, an orchestrator agent or the agent holding the task. Management Square compares these orchestration modes, what each costs in model calls, testing and traceability, and the situations in which each is the right choice.

Read the analysis
An AI centre of excellence: what for, and until when?

An AI centre of excellence is a cross-functional team that pools what several artificial intelligence projects have no interest in redoing separately: scarce skills, a technical foundation, an evaluation method. Management Square examines here the questions an executive committee asks before creating one: its usefulness, its reporting line, what it centralises, the risk of a bottleneck and its lifespan.

Read the analysis
Building the budget for an AI project, from cost items to the choice between build and buy

The budget for an AI project can be broken down into six items: framing, data, build, evaluation, operation and compliance, part of which is spent once and part of which recurs as long as the service exists. Management Square describes each item here by its nature and by what makes it vary, then what changes with the choice between building, adapting an existing model and buying a service.

Read the analysis
Placing an enterprise's AI maturity on observable signs, then choosing the next step

An enterprise's AI maturity can be defined as its ability, established on evidence, to take an artificial intelligence use case from idea to operated service, then to start again with the next one. Management Square proposes here an assessment grid in five dimensions (uses, data, operations, oversight, skills), with the observable signs of each level and the next step they call for.

Read the analysis
The ROI of artificial intelligence is calculated on full costs and observed gains

The ROI of artificial intelligence relates the net gains of an AI project to its full cost, over a period and according to conventions written down alongside the figure. Management Square presents here what can be measured (build and operating costs, observed gains, timescales), how to establish it and what each measure does not say, with a worked example given as a hypothesis.

Read the analysis
Building an enterprise AI roadmap, the steps in order

An enterprise AI roadmap is, by working convention, the document that orders over time the artificial intelligence use cases an organisation has selected, the data workstreams and foundations on which they depend, and the decision points at which each one is continued, reworked or stopped. Management Square proposes an order of construction here, with the document each step delivers and the person accountable for it.

Read the analysis
Docker vs Kubernetes: building containers, orchestrating them, or both

Docker is used to build container images and run them on a machine, Kubernetes to orchestrate containers across a set of machines: the Docker vs Kubernetes comparison concerns two complementary levels. Management Square compares their roles here and describes the situations where one is enough without the other.

Read the analysis
Kubernetes monitoring: the indicators to track, from cluster to application

Kubernetes monitoring is the continuous tracking of the state and consumption of a cluster at four levels: control plane, nodes, workloads and applications. Management Square details here the indicators to track at each level, how to read them and what none of them tells you about the other levels.

Read the analysis
Data provenance vs data lineage, two traces for two requests

Data provenance vs data lineage: provenance describes the origin and history of a piece of data (source, collection conditions, versions, owner), lineage describes its journey through the information system and the transformations it undergoes. Management Square compares the two notions here, whose separation is a matter of convention, and indicates which requests each one answers.

Read the analysis
Which IT migration should you undertake, and when is it better to postpone?

An IT migration transfers an infrastructure, an application or data from a source environment to a target environment, until the source is shut down. Management Square gives here the criteria for choosing the type of migration and the cutover strategy, preparing data migration and rollback, and recognising the cases where it is better to refrain or to postpone.

Read the analysis
Data anonymisation: techniques, limits and choices according to use

Data anonymisation makes it impossible, by reasonable means, to identify the people in a dataset, which then falls outside the scope of the GDPR; a pseudonymised dataset remains personal data for anyone who can trace back to the person. This Management Square article compares the techniques, indicates which one to choose according to use and describes how to verify the result.

Read the analysis
Distributed systems observability begins with a journey traced end to end

Distributed systems observability relies on distributed tracing: an identifier propagated from service to service links the operations of a single request, then the logs and metrics that relate to it. Management Square describes here the method for instrumenting a journey end to end, from context propagation to sampling, with what each step produces and who is responsible for it.

Read the analysis
LLM fine-tuning: should you adjust the model, or stick to the prompt and RAG?

Fine-tuning an LLM continues the training of an existing language model on selected examples in order to change its behaviour lastingly: a format, a register, a narrow task. It brings neither up-to-date knowledge nor verifiable sources: Management Square sets out here the criteria that justify fine-tuning and the cases where the prompt or RAG are enough.

Read the analysis
The difference between AI and generative AI: when to choose a predictive model, when to choose a generative model

The difference between AI and generative AI lies in what the system produces: AI refers to the whole field, including predictive models that estimate a value or classify a case, whereas generative AI, one of its branches, produces new content. Management Square compares the two families here and indicates in which situations each is the right choice.

Read the analysis
A data quality audit traces symptoms back to their causes

A data quality audit is a bounded examination, commissioned by a business department, which measures the state of the data consumed by a use, traces each defect back to its cause and delivers a remediation plan ranked by effort. Management Square describes here the symptoms that justify it, the examinations it carries out in sequence and the decision with which it ends.

Read the analysis
Which lever should you pull first to improve data quality?

Improving data quality means choosing, for each defect whose cause has been established, the lever that stops it from forming again: a fix at source, a data entry rule, a check in the processing pipeline or the appointment of an owner. Management Square sets out here the criteria for this choice, the order in which to pull the levers and the cases where it is better not to correct.

Read the analysis
Data quality testing methods: what each family detects, and what it lets through

Data quality testing methods are automated checks that compare a data set with an explicit expectation: schema, completeness, uniqueness, freshness, consistency between sources, volume and distribution. Each detects one type of defect and lets others through: Management Square compares here these families of tests and their place in data pipelines.

Read the analysis
Cyber crisis management: prepare before, decide during, rebuild after

Cyber crisis management is the organisation that allows an entity to decide, contain an attack and inform when its digital tools are seriously affected and its usual processes are no longer enough. Management Square describes here what each of the three phases of the crisis (before, during, after) produces, who is accountable for it and which regulatory notifications run in parallel.

Read the analysis
Observability vs monitoring: what each one tells you, and at what price

Monitoring checks states defined in advance and alerts when a threshold is crossed; observability makes it possible to query the signals emitted by a system to explain a failure that nobody had anticipated. Management Square specifies here, in the observability vs monitoring debate, when the former is enough and what the switch requires in terms of organisation and costs.

Read the analysis
Retaining several cloud providers: multicloud strategy put to the test of an executive committee

A multicloud strategy deliberately spreads an organisation's uses across several cloud providers, for a named reason and at the price of an accepted complexity. Management Square reviews here the questions senior management would do well to ask before retaining one more provider: rationale, cost, portability, applicable texts and governance.

Read the analysis
Stay on site or move: the trade-offs of a hybrid cloud migration

A hybrid cloud migration moves part of an information system's workloads to a cloud provider and keeps the rest on an infrastructure operated in-house, the two remaining interconnected. Management Square sets out here the criteria that decide the fate of each workload (data, latency, licences, dependencies) and the signs of a hybrid cloud that is only a stage.

Read the analysis
Data governance for AI, seen from the committee that authorises a use case

Data governance for AI applies to the datasets and documents consumed by an AI system the roles and rules that establish who is accountable for a piece of data, who may use it and on what conditions. Management Square reviews here the questions a committee asks before authorising a use case: rights of use, quality, traceability, personal data and responsibilities.

Read the analysis
In what order should AI governance be implemented?

Implementing AI governance is the sequence of steps by which an organisation lists its AI systems, classifies them by risk level, designates who is accountable for them, sets up an approval process and organises their monitoring. Management Square describes these steps here in order, what each one produces and who is accountable for it, in the light of Regulation (EU) 2024/1689.

Read the analysis
DORA and NIS2: what are the differences, and which text applies to a financial entity?

The differences between DORA and NIS2 start with their nature: DORA, Regulation (EU) 2022/2554, applies directly to financial entities, whereas NIS2, Directive (EU) 2022/2555, covers essential and important entities in the sectors of its annexes and operates through national law. Management Square compares the two texts here, article by article, and describes how they fit together for a financial entity.

Read the analysis
Reducing the cloud bill with FinOps, backed by indicators

With FinOps, reducing the cloud bill first requires measuring where spending can fall: share of allocated spend, unit cost, commitment utilisation rate and unused resources. Management Square sets out here the calculation convention for each of these indicators, how to read it and what an isolated figure does not tell you.

Read the analysis
Cloud migration: the mistakes to avoid leave traces in the plan, the cutovers and the bill

In a cloud migration, four framing mistakes are to be avoided: an inventory without the dependencies, a landing platform delivered too late, applications whose fate is undecided, a source environment with no shutdown date. Management Square shows here how to spot them in the wave plan, the cutover log and cost tracking, and what correcting them involves mid-programme.

Read the analysis
Where to start with a data governance strategy

A data governance strategy sets the uses to be served, the order in which data domains are taken on, the people accountable for them, their resources and how the effect obtained is measured. It begins with a single domain, chosen on explicit criteria: Management Square details these criteria here, what the legal texts require and the cases where it is better to postpone.

Read the analysis
A sovereign RAG architecture is judged by where each component runs

A sovereign RAG architecture is a retrieval-augmented generation set-up in which each component (models, vector index, corpus, logs) runs in an environment verified for the place of processing, the law applicable to the provider and control over operations. Management Square indicates where to place each component, according to which criteria, and when this constraint is not justified.

Read the analysis
AI Act obligations for businesses, read alongside their application timetable

AI Act obligations for businesses depend on the role held for each AI system, provider or deployer, and on that system's risk level; their timetable is read in Article 113 of Regulation (EU) 2024/1689, as amended by Regulation (EU) 2026/1744. Management Square reads these provisions here in the consolidated version of 27 July 2026 and draws the consequences for a project.

Read the analysis
Which enterprise AI agent use cases can a committee authorise?

Enterprise AI agent use cases can be classified by what the agent is allowed to do: prepare a file, sort requests, propose an action or execute it within set limits. Management Square sets out the questions a committee asks before authorising an agent: the task, the actions, the permissions, human oversight and measurement.

Read the analysis
A method for securing a cloud infrastructure, and an owner for each step

Securing a cloud infrastructure means protecting what the provider leaves to its customer: identities, resource configuration, encryption keys, logs and backups. Management Square proposes here a method in six ordered steps, with the deliverable each one produces and the person who answers for it.

Read the analysis
Page contents

We write for those who have to decide: launch a project, stop it, choose an architecture, commit a budget. Our insights start from situations encountered on assignments and seek to answer a specific question, rather than to comment on the news.

What you will find here

In-depth analysis of artificial intelligence, data, cloud and platforms. Each article opens with a two-sentence summary and ends with links to the relevant areas of expertise and definitions.

We favour three types of topic: methods that have proven themselves, mistakes we see being repeated and the questions to ask before committing to a project.

Our editorial line

Three rules guide what we publish.

Accuracy first. A regulatory text is cited by its reference. A figure is sourced, or it is not published.

The decision-maker’s point of view. We explain what a choice implies in terms of cost, risk and organisation, not only how a technology works.

Restraint. No promises, no superlatives. If an approach has limits, we write them down.

How we choose our topics

Our topics come from three sources. First, the questions our clients ask us on assignments: when the same question comes up in several organisations, it deserves a written answer. Second, changes in the regulatory framework, because they alter what is possible and how to prove it. Finally, feedback from our consultants on what worked and what did not.

We do not publish to keep to a schedule. An article appears when we have something useful to say on a topic, and it is updated when the context changes.

Further reading

If you are looking for the definition of a term, the glossary is more direct. If you are looking for what we can do for you, see our expertise.

FAQ

Frequently asked questions

Another question? Ask us

What topics do Management Square's insights cover?

This page brings together the firm's analysis of artificial intelligence, data, cloud and platforms. It covers methods, common mistakes and the questions to ask before launching a project.

Where have the old blog articles gone?

The articles from Management Square's previous website have been carried over into this section. The oldest content, devoted to project management and certifications, has been grouped into the site's glossary, in the form of updated entries.

A project to bring to completion?

A first conversation is enough to understand your challenge, identify the right people and tell you frankly what we can bring.