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

Glossary

Glossary of AI, data, cloud and project management

Short, accurate definitions without unnecessary jargon, so that decision-makers and teams are talking about the same thing.

In brief

The Management Square glossary defines the terms of artificial intelligence, data, cloud, software engineering and project management. Each entry gives a short definition, the related notions not to be confused with it and a concrete example.

"Cloud au centre" doctrine The "cloud au centre" (cloud at the centre) doctrine is the French State's doctrine on the use of cloud computing, set by Prime Minister's circular no. 6282-SG of 5 July 2021 and updated by circular no. 6404/SG of 31 May 2023. The cloud au centre doctrine requires a cloud solution to be sought for every new digital project and governs the use of commercial offerings according to data sensitivity. ADKAR model The ADKAR model is a change management model from the company Prosci, which describes change at the level of an individual through five elements: Awareness, Desire, Knowledge, Ability and Reinforcement. Prosci presents the ADKAR model as a model and not as a complete methodology: it serves to identify, for a given population, the first element that is missing. AI Act (European regulation on AI) The AI Act is Regulation (EU) 2024/1689 of 13 June 2024, which lays down harmonised rules for placing artificial intelligence systems on the market and using them in the European Union. It grades obligations according to the level of risk of the use and allocates them between several actors, first and foremost the provider and the deployer. AI agent Agentic AI refers to systems in which an AI agent pursues a goal over several steps: it relies on a language model to plan, call tools, observe the result and adjust its course. An AI agent differs from a chatbot, which answers without acting, and from conventional automation, whose sequence is written in advance. AIOps AIOps (artificial intelligence for IT operations) refers to applying statistical and machine learning methods to IT operations data (metrics, logs, traces, alerts, tickets) to group alerts, detect anomalies and assist diagnosis. A term of common usage, AIOps draws on monitoring and observability data without replacing either. API gateway An API gateway is the component placed in front of a set of services to offer their clients a single entry point: it receives each call, routes it to the relevant service and can provide common functions, such as authentication or rate limiting. The notion comes from common usage and from published architecture patterns. API management API management refers both to the discipline that governs the publication, security, monitoring and retirement of an organisation's programming interfaces, and to the family of tools that supports it. A breakdown in common use distinguishes a gateway, which receives calls, administration functions, a portal for developers and usage monitoring. Architecture decision record (ADR) An architecture decision record (ADR) is a short document that records an architecture decision, the context that motivated it and its consequences. Michael Nygard described the format in a blog post of 15 November 2011; a project's ADRs, numbered and kept even when superseded, form its decision log. BCBS 239 BCBS 239 refers to the "Principles for effective risk data aggregation and risk reporting", a document published in January 2013 by the Basel Committee on Banking Supervision. BCBS 239 sets out fourteen principles on governance, risk data aggregation, risk reporting and supervisory review, addressed first and foremost to systemically important banks. BCP and DRP The expression IT BCP and DRP refers to two complementary plans. The business continuity plan (BCP) organises how essential activities are maintained during a disruption of the information system; the disaster recovery plan (DRP) organises how that system is brought back into service after an incident that has interrupted it. Behaviour-driven development Behaviour-driven development (BDD) is a software development approach in which the business, developers and testers describe the expected behaviour of an application through concrete examples, written as scenarios in the vocabulary of the business. These scenarios serve as acceptance criteria and can run as automated tests. C4 model Created by Simon Brown, the C4 model describes the architecture of a software system through nested diagrams at four levels: the system context, its containers, their components and the code. This approach to visualisation imposes neither a graphical notation nor a tool. Chief data officer The chief data officer (CDO), or head of data, is the executive responsible, at the leadership level of an organisation, for the framework within which data is defined, governed and put to work for the business. The title is a matter of usage: its scope and reporting line vary from one organisation to another. Circuit breaker The circuit breaker is a software resilience pattern: an object wraps a call to a remote service, counts its failures and, beyond a threshold, immediately returns an error without attempting the call. After a delay, a trial call decides whether to return to normal. Martin Fowler, who describes it in an article dated 6 March 2014, credits Michael Nygard's book "Release It!" with popularising it. Clean architecture Clean architecture is a model for organising code proposed by Robert C. Martin, which arranges software in concentric circles, from business rules at the centre to technical tools at the edge. It rests on one rule, known as the dependency rule: source code dependencies point only inwards. Cloud Act The Cloud Act (Clarifying Lawful Overseas Use of Data Act) is a United States federal law enacted on 23 March 2018. It provides that a provider of electronic communication or remote computing services must disclose, under the conditions of US law, the data in its possession, custody or control, wherever that data is stored. It also provides for agreements with other States. Complex event processing (CEP) Complex event processing (CEP) detects, in a stream of events, patterns described in advance: a sequence, a repetition or an absence of events within a given time. Each recognised pattern produces a so-called complex event, that is a composite event, which signals a situation to be handled. Composable architecture Composable architecture refers, in market usage, to an information system or a digital platform assembled from independent components, exposed through APIs and replaceable one by one. The expression belongs to commercial usage: its outlines vary according to the software vendors and professional associations that use it, including the MACH Alliance. Cyber Resilience Act The Cyber Resilience Act is Regulation (EU) 2024/2847 of 23 October 2024, which sets cybersecurity requirements for products with digital elements made available on the Union market. Its Article 71 makes it applicable from 11 December 2027, with Article 14 applying from 11 September 2026 and Chapter IV from 11 June 2026. Data Act The Data Act is Regulation (EU) 2023/2854 of 13 December 2023, which sets harmonised rules on access to and use of data in the European Union. Applicable from 12 September 2025, except for dates specific to certain obligations, it covers in particular connected product data, its sharing between businesses and switching cloud providers. Data catalogue A data catalogue is the inventory of an organisation's datasets, described by their metadata: content, structure, owner, classification and access conditions. A term of common usage, the data catalogue serves to find data, understand it and know who is accountable for it, without hosting the data itself. Data contract A data contract is the agreement through which the producer of a dataset declares to its consumers what it provides: structure, meaning of fields, quality rules, service levels and contacts. In practice the data contract takes the form of a versioned file that tools verify, for example in the open Open Data Contract Standard format. Data lake A data lake is a centralised storage space that keeps data of any kind (database extracts, files, logs, images) in its original format, without imposing a structure on it at entry. The structure is applied at read time, by the use that needs it. Data lineage Data lineage is the description of the path a piece of data takes through the information system: the datasets a process reads, the transformations it applies and those it writes, from the system that produces the data to the uses that consume it. A term of common usage, data lineage is represented by a dependency graph, read upstream or downstream. Data mesh The data mesh is an organisational model for analytical data that entrusts responsibility for data to the business domains that produce it, instead of a single central team. It rests on four principles: domain ownership, data treated as a product, a self-service platform and federated governance. Data observability Data observability is the continuous monitoring of datasets based on signals collected on each table: freshness, volume, schema, distribution of values and dependencies. A concept that tool vendors word in different ways, data observability is used to detect an anomaly that no written rule targeted, then to look for its origin. Data product A data product is a set of data that a team makes available to other teams as a product: findable, documented, backed by quality commitments and carried by an identified owner. A term of common usage, the data product is the unit that the data mesh asks each business domain to publish. Data storytelling Data storytelling refers to presenting the result of an analysis in the form of a narrative built for a specific audience, which combines data, visualisations and commentary. A term of common usage, data storytelling aims to make a quantified finding understood, along with the decision it calls for. DevOps DevOps is a set of principles and practices that bring software development (Dev) and IT operations (Ops) closer together, in order to deliver smaller, more frequent and better controlled changes. The term DevOps became established through usage, with no single definition: it covers shared responsibility for the service, automation of the delivery pipeline and measurement of its results. DevSecOps DevSecOps (development, security, operations) refers to building security into DevOps practices and the delivery pipeline: controls run on every change, and the team that delivers handles their results itself. The term DevSecOps comes from usage; OWASP and NIST publish guides that describe how to implement it. Digital product passport The digital product passport, as defined by Regulation (EU) 2024/1781 of 13 June 2024, is a set of data specific to a product, accessible electronically by means of a data carrier present on the product, its packaging or its documentation. This regulation makes the passport mandatory for a product group only through a delegated act of the Commission. DORA Regulation The DORA Regulation (Digital Operational Resilience Act) is Regulation (EU) 2022/2554 on digital operational resilience for the financial sector, applicable since 17 January 2025. It requires financial entities to adopt a common framework for managing IT risk, reporting incidents, testing resilience and controlling their digital service providers. Enterprise architecture Enterprise architecture is the discipline that describes an organisation as a whole, from its objectives to its technical infrastructure, in order to guide its transformations and keep them consistent. The French term architecture d'entreprise corresponds to it; frameworks such as TOGAF and modelling languages such as ArchiMate are attached to it. Error budget The error budget measures the gap between 100% and a service's service level objective (SLO): the amount of failure that this objective allows over its measurement window. In the SRE books published by Google, a written policy sets what teams do when the error budget is consumed, for example freezing production releases until the service is back within its objective. EUCS EUCS (European Cybersecurity Certification Scheme for Cloud Services) is the draft European cybersecurity certification scheme for cloud services that ENISA is preparing under Regulation (EU) 2019/881, known as the Cybersecurity Act. As at 12 November 2026, EUCS has not been adopted: it remains a candidate scheme, with no effect until the European Commission adopts it by implementing act. European Health Data Space The European Health Data Space (EHDS) is the framework established by Regulation (EU) 2025/327 of 11 February 2025 to facilitate access to electronic health data, for care (primary use) and for purposes such as research or statistics (secondary use). Under its Article 105, the regulation applies from 26 March 2027, then in stages until 2035. Event-driven architecture (EDA) Event-driven architecture makes applications communicate through events: one system publishes the fact that a change has occurred, and those interested in that fact react to it without the emitter having to know them. The expression covers several models, which Martin Fowler distinguished in an article of 7 February 2017. Feature flags (feature flipping) Feature flipping consists of placing a feature behind a switch (a feature flag or feature toggle), which makes it possible to turn it on or off at runtime without changing the code. Feature flags thus separate the release of a feature from the deployment of the code that carries it, at the cost of temporary switches that must then be removed. Feature store A feature store is the component of a machine learning pipeline that records the definition of models' input variables (features), stores their values and serves them for training as well as in production. The open source projects Feast and Hopsworks describe the feature store around two modes of access: offline over the history, online with low latency. FinOps FinOps is a cloud financial management discipline that brings technical teams, finance and the business together so that every expense is visible, assigned to an owner and justified by the value it produces. Its practice follows a cycle of three phases repeated continuously: inform, optimise, operate. GitOps GitOps is a set of principles for operating software systems: the desired state of the system is described declaratively and kept in versioned storage, such as a Git repository, then software agents pull it and apply it continuously. The OpenGitOps project, overseen by a working group of the CNCF SIG App Delivery, publishes four principles in a version 1.0.0. Golden path A golden path is the recommended and supported route that an organisation offers its development teams for a common need, such as creating a service or a data pipeline. Spotify's engineering article on the golden path defines it as the "opinionated and supported" path: its tool choices have already been made and it comes with support. Hexagonal architecture (ports and adapters) Hexagonal architecture, also called ports and adapters, is a design pattern described by Alistair Cockburn that separates the core of an application from what surrounds it: interface, database, other systems. The application exchanges with the outside through ports, defined by the purpose of the exchange, and each technology plugs into them through a replaceable adapter. Immutable backup An immutable backup is a copy of data which, once written, can be neither modified nor deleted before the end of a retention period set in advance, including by an administrator account. It preserves a restore point when production and ordinary backups have been encrypted or erased. Infrastructure as code Infrastructure as code (IaC) consists of describing an IT infrastructure (networks, servers, databases, access rights) in versioned files, which a tool applies to create or modify the resources. A term of common usage, infrastructure as code replaces manual operations with changes that can be reviewed, traced and replayed. Internal developer portal An internal developer portal is the web interface where an organisation's development teams find the catalogue of software components, their owners and their documentation, create projects from templates and access the capabilities of the internal platform. The expression is one of common usage: the CNCF "Platforms" white paper speaks of web portals. ISO/IEC 42001 ISO 42001 is the common name of the international standard ISO/IEC 42001:2023, which specifies the requirements and guidance for establishing, implementing, maintaining and continually improving an artificial intelligence management system. Published in December 2023, it is aimed at any organisation that provides or uses products or services that use AI systems. IT landscape mapping IT landscape mapping is the organised representation of the elements of an information system and of their links: processes, functions, applications, data, flows and infrastructure. In IT urbanisation practice, it is presented as four layered views (business, functional, application, technical) and serves to inform decisions, provided it is kept up to date. IT master plan The IT master plan is the document in which an organisation sets, for several years, the direction in which its information system will evolve, the projects to be launched, their order and the resources allocated to them. It starts from the organisation's strategy, a current-state assessment and a gathering of needs, and it is approved by senior management. IT urbanisation IT urbanisation is the approach that organises the gradual transformation of an information system according to an overall plan, by analogy with a city's urban planning: partitioning into zones, districts and blocks, exchange rules between them, a target and the steps to reach it. The term is French usage; international publications speak of enterprise architecture, a broader notion. IT/OT convergence IT OT convergence refers to the coming together of business information systems (IT, information technology) and the industrial systems that monitor or control physical equipment (OT, operational technology). IT OT convergence moves data from installations to business systems; in return it reduces the isolation of those installations, hence the continued use of segmented networks. ITIL ITIL is a framework of good practice for IT service management (ITSM): it describes how to design, deliver, operate and improve the services an IT department provides to its users. It does not deal with running projects, but with what happens once their result has been put into service. Lakehouse The lakehouse is a data platform architecture that combines, on a single foundation, the open and loosely constrained storage of the data lake with the guarantees of the data warehouse: reliable transactions, controlled schemas and high-performance queries. It aims to serve BI, data science and machine learning from a single copy of the data. Landing zone A cloud landing zone is the foundation an organisation sets up with a cloud provider before deploying its applications there: account hierarchy, identities and access, network, security controls, logging. The term comes from the usage of cloud providers, whose documentation gives similar definitions. Lean portfolio management (LPM) Lean portfolio management (LPM) is, in the SAFe framework, the discipline that aligns strategy and execution by applying lean and systems thinking approaches to investment funding, portfolio operations and governance. According to the framework's website, a portfolio's budget is allocated to value streams, not to projects. LLM (large language model) An LLM (large language model) is a neural network trained on very large volumes of text to predict the continuation of a sequence, which enables it to write, summarise, translate or produce code. It generates the most plausible text, not verified text, hence the phenomenon of hallucinations. LLMOps LLMOps (large language model operations) refers to the practices and tools used to put into production, evaluate, monitor and evolve an application built on a large language model. A term of common usage, whose scope varies from one source to another, LLMOps extends MLOps and covers in particular prompt management, evaluation, monitoring in production and cost control. Master data management Master data management (MDM) refers to the set of rules, processes and tools by which an organisation maintains a single, shared version of the entities used by several applications, such as customers, products or suppliers. The term is one of common usage, and its wording varies from one source to another. Matrix organisation The matrix organisation is a structure in which each employee reports both to a functional manager and to a project manager. It exists in three variants (weak, balanced and strong matrix) depending on the authority granted to the project manager, and sits between the functional organisation, built on disciplines, and the project-based organisation, built on dedicated teams. MCP protocol (Model Context Protocol) The MCP protocol (Model Context Protocol) is an open standard that describes how an AI application connects to external systems: data sources, tools, prompt templates. The MCP protocol rests on a client-server architecture and on JSON-RPC 2.0 messages; it does not govern how the application uses the language model. Micro frontend A micro frontend is a part of a web application's interface, developed and delivered separately by a team, then assembled with the others to form a single site in the user's eyes. Cam Jackson, in a martinfowler.com article dated 19 June 2019, speaks of an architectural style where independently deliverable frontend applications are composed into a greater whole. Microservices architecture Microservices architecture is a software architecture style that builds an application as a set of small services, each running in its own process, organised around business capabilities and independently deployable. James Lewis and Martin Fowler, in their article of 25 March 2014, state that there is no formal definition of it and describe common characteristics. MLOps MLOps (machine learning operations) is the set of practices, roles and tools that make it possible to put machine learning models into production and keep them there reliably. It extends the principles of DevOps to two objects that conventional software does not have: the data and the trained model. MOA, MOE, AMOA MOA, MOE and AMOA name three roles in an information system project: the maîtrise d'ouvrage (project owner), on whose behalf the system is built, the maîtrise d'œuvre (project contractor), which designs and builds it, and the assistance à maîtrise d'ouvrage (business-side project support), which supports the first without deciding for it. The first two come from construction; their use in IT is a matter of custom. Modern data stack The modern data stack is a family of analytical data architectures built around a cloud data warehouse: data is loaded raw, then transformed in place, by specialised tools assembled layer by layer. A term of usage, the modern data stack refers to a way of assembling, which each organisation puts together with its own tools. mTLS (mutual TLS authentication) mTLS (mutual TLS, mutual TLS authentication) is a TLS connection in which the client also presents a certificate and proves that it holds the private key, in addition to the usual authentication of the server. mTLS authenticates both ends of a connection, without saying what the client is authorised to do, and requires managing the certificate life cycle. Multi-agent system A multi-agent system is an architecture in which several AI agents, each with its own instructions, tools and context, cooperate on the same task, for example coordinated by an orchestrator agent. Compared with a single tool-equipped agent and with a chain of calls whose steps are fixed by code, a multi-agent system adds exchanges to coordinate, tokens to pay for and trajectories to verify. Mutation testing Mutation testing is a technique for evaluating an automated test suite: a tool introduces small faults into the code, one at a time, and checks that at least one test fails for each. The share of faults detected, called the mutation score, indicates how well the tests can catch a defect, which code coverage does not measure. MVP (minimum viable product) A product MVP (minimum viable product) is the version of a new product reduced to what is needed to test a hypothesis with real users, before investing further. The product MVP differs from the proof of concept, which tests feasibility, from the prototype, which shows without delivering the service, and from the pilot, which tests the rollout of a solution already chosen. NIS2 Directive The NIS2 Directive is Directive (EU) 2022/2555 of 14 December 2022, which aims at a high common level of cybersecurity across the European Union. It requires essential and important entities in critical sectors to adopt risk management measures and to notify significant incidents, under rules that each Member State transposes into its own law. OKR (objectives and key results) An OKR (Objectives and Key Results) combines an objective, expressed in words, with a few measurable results that make it possible to check, at the end of a period, whether it has been achieved. The term also refers to the management practice of setting, publishing and scoring such objectives at a regular cadence. OpenTelemetry OpenTelemetry is an open source project of the Cloud Native Computing Foundation (CNCF) that provides specifications, programming interfaces, software development kits and a collector to generate, collect and export telemetry data: traces, metrics and logs. Vendor-neutral, OpenTelemetry is not an observability backend: it neither stores nor displays this data. PI planning PI planning is the SAFe framework's fixed-cadence event that brings together all the teams of an Agile Release Train to align them on a shared mission and vision. According to the framework's website, it takes place every 8 to 12 weeks, usually over 2 days, and results in objectives to which the teams commit. Platform engineering Platform engineering is the discipline that designs and evolves an internal platform giving development teams, on a self-service basis, the means to deliver and operate their applications. The platform is managed as a product whose users are the organisation's developers. PMI certifications PMI certifications are the professional credentials awarded by the Project Management Institute to attest a person's knowledge and experience in project, programme or portfolio management. The best known are the PMP for project management, the PgMP for programme management and the PfMP for portfolio management. PMO (Project Management Office) The PMO (Project Management Office) is the structure that defines the rules for managing an organisation's projects, consolidates their progress and prepares management's trade-off decisions. By extension, the acronym also refers to the person who performs this function. PPM (project portfolio management) PPM (Project Portfolio Management) is the discipline of choosing, prioritising and tracking all of an organisation's projects according to its strategy and its actual capacity. It answers a question that project management does not ask: are we doing the right projects? PRINCE2 PRINCE2 (PRojects IN Controlled Environments) is a structured project management method, born in the UK civil service, which organises a project into stages decided one by one by a project board. It rests on seven principles, seven themes (called practices since version 7) and seven processes, and differs from the PMBOK, which is a body of knowledge guide, and from the PMP, which is a certification. Product owner In Scrum, the product owner is the person accountable for maximising the value of the product resulting from the team's work, and for managing the Product Backlog. The November 2020 Scrum Guide makes it one of the three accountabilities of the Scrum Team, alongside the Developers and the Scrum Master. Project-based organisation The project-based organisation, known as projectized in project management literature, is a structure in which teams are formed around projects and placed under the direct line authority of the project manager, who holds the budget and resources. It offers speed and single accountability, at the cost of duplicated skills and uncertainty for teams at the end of each project. RAG (retrieval-augmented generation) RAG (retrieval-augmented generation) is an architecture in which a language model writes its answer from passages retrieved beforehand from a document base. It allows answers to draw on a private, up-to-date corpus, with verifiable sources, without retraining the model. Refactoring Refactoring is the modification of a piece of software's internal structure without changing its observable behaviour, in order to make it easier to understand and cheaper to modify. It proceeds through a series of small transformations, with the software remaining in working order after each one. RTO and RPO The RTO RPO pair refers to the two objectives of a disaster recovery plan. The RTO (recovery time objective) is the maximum tolerable period of interruption of an activity or a service; the RPO (recovery point objective) is the maximum tolerable data loss, expressed as a period of time. SAFe (Scaled Agile Framework) SAFe (Scaled Agile Framework) is a framework that describes how to apply agile and lean principles when several teams, sometimes several hundred people, develop the same product or the same system. It relies on agile release trains that plan together, at a fixed cadence, during an event called PI planning. SecNumCloud SecNumCloud is ANSSI's qualification framework for cloud service providers. It sets the technical, organisational and legal requirements a service must meet to be qualified, and serves as a benchmark for organisations that host sensitive data or processing. Security accreditation (homologation de sécurité) Security accreditation (homologation de sécurité) is, under Article 4-3 of French Decree No. 2019-1088 of 25 October 2019, a formal decision attesting that security risks have been identified, that the necessary measures are implemented and that residual risks are accepted. This article requires security accreditation before the components of the State information system are put into service, unless exempted. Self-service BI Self-service BI refers to a way of organising business intelligence in which business users build their own reports and analyses, without ordering them from the central BI team. A term from common usage, self-service BI is practised within a framework known to the organisation: tools provided, prepared data and publication rules. Semantic layer The semantic layer is the software layer placed between the data in a warehouse and the analysis tools, which holds the shared definitions of metrics and dimensions. In a semantic layer, a metric is calculated in a single place and returns the same value to every tool that queries it. Shadow AI Shadow AI refers to the use of artificial intelligence tools by members of an organisation without its framework or approval. This term from common usage, formed by analogy with shadow IT, describes the situation of a use within a given organisation, not a category of tools. Showback and chargeback Showback shows a team the cost of the IT services it consumes, without any accounting entry. Chargeback (internal recharging) charges that cost to its budget, in the organisation's financial systems. According to the FinOps Foundation framework, showback and chargeback differ by this accounting formality, and neither is more mature than the other. SLO, SLI and SLA The SLO SLI SLA triad refers to three related notions in managing the reliability of a service. The SLI (service level indicator) is a measure of the level of service delivered, the SLO (service level objective) the target value set for that measure, and the SLA (service level agreement) a contract with users that attaches consequences to meeting or missing the objectives. SRE (Site Reliability Engineering) SRE (Site Reliability Engineering) is a discipline that treats the operation of IT systems as a software engineering problem. It manages reliability through measured objectives (SLOs) and an error budget that arbitrates between shipping new features and keeping the service stable. Strangler fig pattern The strangler fig pattern is a way of replacing a legacy system in stages: a new system is built around the old one and takes over its functions one by one, until the old one can be switched off. The name comes from a metaphor by Martin Fowler, drawn from a fig that grows on a host tree. Technical debt Technical debt refers to the future extra cost that a team accepts, deliberately or not, when it chooses a quick or imperfect software solution rather than a lasting one. Like financial debt, it bears interest: every change to the software costs more for as long as the original flaw has not been corrected. Test pyramid The test pyramid is a model for distributing automated tests by level: many unit tests at the base, fewer tests as their scope widens, very few end-to-end tests at the top. The articles by Martin Fowler and Ham Vocke, on martinfowler.com, trace it back to Mike Cohn's book "Succeeding with Agile". Third-party application maintenance (TMA) Third-party application maintenance (TMA, from the French tierce maintenance applicative) refers, in French usage, to the maintenance of one or more applications entrusted by an organisation to an external provider. Its scope, set by the contract, can combine the correction of defects, functional changes, adapting the application to changes in its environment and preventive work. TOGAF TOGAF is a standard from The Open Group that offers a method and a framework for conducting an enterprise architecture approach. It is organised around an architecture development method, the ADM (Architecture Development Method), and is presented as a body of material to be adapted to each organisation's practice. Trusted cloud Trusted cloud (cloud de confiance) is an expression from the French national cloud strategy, announced on 17 May 2021. It refers to cloud services that give data dual protection, technical and legal: verified IT security and the exclusive application of European law. Later official communications base it on the ANSSI SecNumCloud qualification. Vector database A vector database is a system that stores numerical vectors, called embeddings, and retrieves those closest to a query vector according to a distance measure. A term of common usage, the vector database is used for similarity search over texts, images or other content, by means of so-called approximate indexes. Zero trust Zero trust is a cybersecurity model in which no trust is granted to a user or a device solely because of its location on the network: every access to a resource is authenticated, authorised and reassessed. The NIST SP 800-207 publication describes zero trust as a set of guiding principles, not as a single architecture.
Page contents

A project often fails for a simple reason: the people around the table do not understand the same thing by the same word. PMO, RAG, FinOps, data mesh, SRE: these acronyms circulate in committees without their meaning always being shared.

Why a glossary

This glossary has a modest but useful ambition: to give each term a short, accurate definition that a non-specialist can read. It covers the terms that come up in AI, data, cloud and project management projects.

Each entry follows the same structure. It opens with a two-sentence definition that you can reuse as it stands in a note or a presentation. It then explains what the notion is for, what its variants are and which related terms it is confused with. It ends with a concrete example.

The domains covered

The entries are divided between our five areas of expertise.

  • Project management: project office, portfolio management, frameworks and certifications, forms of organisation.
  • Artificial intelligence: language models, retrieval-augmented generation, agents, European regulatory framework.
  • Data: platform architectures and model industrialisation.
  • Cloud and platforms: reliability, cost control, platform engineering, texts on resilience and security.
  • Software engineering: technical debt.

How we handle regulatory texts

Entries devoted to a regulation or a directive cite the text by its official reference. They describe its purpose and its main principles, without replacing a legal analysis of your situation. Precise deadlines and obligations depend on your business: check them with your legal department or the competent authority.

Further reading

Each entry links to the related terms and to the corresponding service among our expertise. For more in-depth analysis, see our insights.

FAQ

Frequently asked questions

Another question? Ask us

Who is this glossary for?

This glossary is for decision-makers, project managers and business teams who come across technical terms without always having a clear definition of them. Each entry can be read in a few minutes and assumes no prior knowledge.

How are the definitions written?

Each entry opens with a two-sentence definition, then explains what the notion is for, its variants, the related terms not to be confused with it and a concrete example. Regulatory texts are cited by their official reference.

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.