AI, compliance and delivery glossary by Access International

AI systems that pursue multi-step goals autonomously by reasoning, planning, calling tools, and adapting to feedback, rather than answering single-turn queries. Combines a foundation model with tool use (web search, code execution, file operations, API calls) and memory. Notable examples: AutoGPT and successors, Anthropic Claude with tool use, OpenAI Operator, Microsoft Copilot Studio agents. Raises new compliance questions around autonomy, traceability, and accountability.

Google search feature launched May 2024 in the US (rolled out globally through 2025) presenting a generative AI summary above traditional search results for selected queries. Overviews synthesize information from indexed pages and cite their sources. Significant impact on SEO: pages cited in Overviews are not always top-ranked classically, opening differentiation space. Requires strong content quality, clear structure, and citable passages to be selected.

Category of search interfaces using generative AI to deliver direct answers rather than (or alongside) traditional ranked links. Major players: Google AI Overviews, Microsoft Bing Copilot, Perplexity, You.com, Anthropic Claude with web search, ChatGPT search. By 2026, AI search captures a significant share of B2B and B2C information queries, fundamentally changing how users discover content and brands.

Tiered customer retention model used in subscription industries (telecom, banking, SaaS) where escalating retention actions are triggered by predictive churn signals: light touch (no-cost service improvement), medium touch (personalized offer or coaching), high touch (advisor outreach with significant commercial gesture). Designed to spend retention budget where it preserves the highest customer lifetime value, not at the gates indiscriminately.

Capitalized lessons from ATLAS-led modernization projects, currently 56 entries spanning COBOL, Delphi, BizTalk, and robotics. Each learning captures a recurring pattern, a counter-intuitive trap, or a productivity insight. The learnings library is the living memory of the methodology: it grows with each project and is consulted at the start of new engagements.

Access International's proprietary framework for industrialized legacy modernization. Structures COBOL, Delphi, BizTalk, PowerBuilder, AS/400, mainframe, and robotics migrations toward modern stacks (.NET Core, TypeScript, Java, Python, cloud) with a single objective: proven functional parity with controlled regression risk. Articulated around 10 steps, 10 guiding principles, 56 capitalized learnings, 9 typical discrepancies, and 19 known traps. Combines vibe coding and methodological rigor.

Application of the ATLAS methodology to the multi-brand industrial robotics world. Covers KUKA KRL, ABB RAPID, Universal Robots URScript, Fanuc Karel, and Siemens Ladder/SCL. Five services: robot code archeology, AI-assisted trajectory optimization in simulator, safety practices audit, data twin generation from code, and auto-generated multilingual documentation. No cross-brand migration, no safety recertification, no production deployment.

Four largest professional services networks: Deloitte, PwC, EY, KPMG. Originally audit and tax firms, all four developed substantial consulting and technology services arms over the 2000s-2020s. In IT services, Big Four firms compete with tier-1 integrators on transformation programs and frequently partner on regulated industries (financial services, public sector, healthcare).

Engagement model where the provider sets up and operates a dedicated team or capability (typically nearshore or offshore) for a contracted period, then transfers ownership to the client. Common for clients establishing captive operations abroad without taking the early-stage execution risk. Tunisia is a recurring BOT destination for French and Canadian groups.

Dedicated organizational structure (in-house, with provider, or hybrid) that pools expertise on a domain (Salesforce, ServiceNow, AI, data, security) and serves multiple business units across the enterprise. Key benefit: avoids per-project reinvention and accelerates capability spread. Provider equivalent to French CDC (Centre de Compétences).

Systems managing the end-to-end contract lifecycle: drafting, negotiation, electronic signature, execution, renewal, archiving. Major vendors: Ironclad, Icertis, DocuSign CLM, Conga, Agiloft, SirionLabs. Particularly valuable for legal departments managing hundreds to thousands of active contracts with monitoring of renewals, risk clauses, and obligations.

Central repository of IT assets and their relationships (servers, applications, services, users, locations, contracts). The single source of truth for ITSM and ITOM operations. Quality of CMDB data is the most common bottleneck in ITSM maturity: a poorly maintained CMDB undermines incident triage, impact analysis, and change management.

System managing maintenance of industrial and clinical equipment: asset registry, preventive and corrective maintenance scheduling, spare parts management, cost tracking, failure history. Foundation for modern predictive maintenance via IoT and AI. Major vendors: IBM Maximo, SAP PM, Infor EAM, Fiix, UpKeep. Healthcare-specific: Asset Plus, Optimus Biomed.

Delivery model where Access International acts as a force multiplier alongside a tier-1 prime contractor (CGI, Capgemini, Cofomo, Accenture, etc.) rather than as a direct prime. Brings ATLAS methodology, vibe coding capacity, and nearshore Tunisia economics under the prime's contractual umbrella and methodology. Frequent for large public sector or enterprise programs where prime contractor selection is fixed.

First comprehensive state-level US AI law, signed May 2024, effective 1 February 2026. Targets developers and deployers of high-risk AI systems making consequential decisions (employment, education, financial services, health, housing, insurance, legal services). Imposes duties of reasonable care to avoid algorithmic discrimination, risk management programs, impact assessments, transparency notices to consumers, and reporting to the Colorado Attorney General.

ServiceNow product extending the platform to customer-facing service operations: case management, agent workspace, knowledge bases, omnichannel routing, customer portals, field service integration. Increasingly combined with Now Assist (ServiceNow's generative AI suite) for case summarization, response drafting, and self-service.

Directive (EU) 2022/2464 mandating detailed sustainability reporting (ESG) by EU large undertakings and listed SMEs, replacing the NFRD. Reporting follows the European Sustainability Reporting Standards (ESRS), is audited (limited then reasonable assurance), and published in iXBRL format. Phased application from 2025 fiscal year reporting through 2028. The 2025 Omnibus package adjusts thresholds and timeline. Has extraterritorial effect on non-EU groups with significant EU revenue.

Lightweight digital representation of a physical asset, focused on data structures (kinematics, sensors, I/O maps, expected states) without the heavyweight physics simulation of a full digital twin. Useful as a foundation layer for predictive analytics, anomaly detection, and asset documentation. Often extracted from existing legacy code rather than built from scratch.

Microsoft's AI-powered ambient clinical documentation solution: records the consultation, generates the structured medical note, suggests ICD-10 codes. Built on Nuance technology (acquired by Microsoft in 2022) and integrated with Epic, Cerner, and other major EHRs. Deployed in 150+ health systems globally. Reference benchmark for AI clinical documentation in the US market.

ServiceNow ITOM module that automatically discovers IT infrastructure (servers, network devices, storage, cloud resources, applications) and populates the CMDB. Uses agentless probes (SSH, WMI, SNMP, cloud APIs) to inventory and detect changes. Reduces manual CMDB maintenance from continuous overhead to exception management.

ATLAS deliverable: structured register of every observed difference between legacy and modernized output (encoding, rounding, date handling, operation order, edge cases). Each entry is documented, categorized, validated as acceptable, or scheduled for correction. Becomes a contractual artifact between client and provider. Auditable. Often required for regulated migrations (banking, insurance, healthcare).

Regulation (EU) 2022/2554 strengthening operational resilience of the EU financial sector, fully applicable since 17 January 2025. Covers banks, insurers, investment firms, crypto-asset service providers, and their critical ICT third-party providers (cloud, SaaS). Five pillars: ICT risk management, incident reporting, digital operational resilience testing (TLPT), third-party risk management, information sharing. Often combined with NIS 2 and EU AI Act for financial AI use cases.

French term for IT services firm, replacing the older SSII (Société de Services en Ingénierie Informatique). Mostly synonymous with ITSF but carries connotations specific to the French market: a recognized industry with public statistics (Numeum association), distinct labor categories (cadre informatique), and a dominant project-based / time-and-materials model. Access International is an ESN headquartered in Tunisia, serving the French and broader EU market.

The European Union's Artificial Intelligence Act, the world's first comprehensive AI regulation, adopted March 2024 and published in the Official Journal on 12 July 2024. It uses a risk-based approach: unacceptable risk (banned), high-risk (Annex III, strict obligations), limited risk (transparency), minimal risk (free use). Applies extraterritorially: any provider or deployer placing AI on the EU market or affecting EU persons is concerned, regardless of where the company is based.

Term coined by Stanford in 2021 referring to large neural network models trained on broad data at scale and adaptable to many downstream tasks. Includes LLMs (GPT, Claude, Gemini, Llama, Mistral), vision models, multimodal models, and code models. The shift from task-specific models to foundation models is the defining technology trend of the 2020s and underpins generative AI, RAG, and agentic systems.

ATLAS principle: the new system must produce an output identical (down to the bit, where applicable) to the legacy system on a representative test dataset before any optimization or feature change. Modernization is a capability transfer with fidelity commitment, not a creative redesign. Optimization happens after parity is proven.

Regulation (EU) 2016/679 protecting personal data of EU residents, applicable since 25 May 2018. Establishes principles (lawfulness, purpose limitation, data minimization, accuracy, storage limitation, integrity, accountability), data subject rights (access, rectification, erasure, portability, objection), and obligations on controllers and processors. Sanctions up to €20M or 4% of worldwide annual turnover. Extraterritorial: applies to any organization processing EU residents' data, regardless of location.

GDPR article prohibiting in principle the processing of sensitive personal data: racial or ethnic origin, political opinions, religious or philosophical beliefs, trade union membership, genetic and biometric data, health data, sex life or sexual orientation. Limited exceptions (explicit consent, employment law, vital interests, public health, scientific research). Health and HR are the most exposed sectors.

Class of AI systems that generate new content (text, images, audio, video, code) from training data, typically based on transformer architectures. Major models include GPT-4 / GPT-5 (OpenAI), Claude (Anthropic), Gemini (Google), Llama (Meta), Mistral. The 2022-2026 wave triggered by ChatGPT redefined enterprise AI use cases from prediction to creation, content augmentation, and conversational interfaces.

Emerging discipline maximizing brand visibility in AI search engine answers (ChatGPT, Perplexity, Claude, Google AI Overviews, Bing Copilot). Distinct from classical SEO: GEO targets citation by LLMs rather than ranking on Google. Levers: self-contained citable passages (50 to 150 words), enriched Schema.org markup (FAQPage, DefinedTerm, Article), llms.txt at site root, dated authoritative sources, hierarchical content structure. Coexists with SEO; does not replace it.

Family of tools covering enterprise governance, risk management, and compliance operations: GDPR records, incident management, internal audits, SOX controls, risk mapping, policy management. Major vendors: ServiceNow GRC, OneTrust, RSA Archer, MetricStream, AuditBoard, LogicGate. Foundational for compliance, legal, and internal audit functions in regulated industries.

Category of the EU AI Act covering AI systems with significant impact on health, safety, or fundamental rights: biometric identification, critical infrastructure, education, employment (recruitment, evaluation), access to essential services (credit scoring, insurance, social benefits), law enforcement, migration, justice. Strict obligations: risk management, data governance, technical documentation, automatic logging, human oversight (HITL), accuracy and cybersecurity, conformity assessment, CE marking. Applicable by 2 August 2027 at the latest.

United States federal law enacted in 1996 protecting the privacy and security of Protected Health Information (PHI). Composed of the Privacy Rule, the Security Rule (administrative, physical, technical safeguards), the Breach Notification Rule, and the Enforcement Rule. Enforced by HHS Office for Civil Rights (OCR). Applies to covered entities (healthcare providers, health plans, clearinghouses) and their business associates. Civil penalties up to USD 2.1M per violation category per year, criminal penalties up to USD 250,000 and 10 years imprisonment for willful violations.

AI architecture pattern where a human validates, adjusts, or supervises AI-generated decisions before they have effect on a user, patient, customer, or employee. Three modes: pre-decision (AI proposes, human approves), post-decision (AI acts, human audits sample), and hybrid (confidence threshold routes high-stakes cases to humans). Explicitly required by EU AI Act Article 14 for high-risk systems. Central reference in NIST AI RMF and ISO 42001 for trustworthy AI.

Industry-recognized security and privacy framework created by HITRUST Alliance, widely adopted by US healthcare providers, payers, and health technology vendors. Maps to HIPAA, NIST 800-53, ISO 27001, PCI DSS, GDPR, and others. Three certification levels: e1 (foundational, 1 year validity), i1 (implementation, 1 year), r2 (risk-based, 2 years). Recognized by major US health systems as a baseline trust signal.

ServiceNow product extending the platform to HR operations: employee case management, lifecycle workflows (onboarding, offboarding, transfers), HR knowledge, HR agent workspace, document generation. Often replaces fragmented HR ticketing tools and integrates with Workday, SAP SuccessFactors, or Oracle HCM.

International standard specifying requirements for an Information Security Management System (ISMS). Latest revision published October 2022, replacing the 2013 version with updated controls reflecting modern threats (cloud, threat intelligence, data masking). Certification through accredited bodies (BSI, DNV, TÜV, AFNOR, etc.) for three years with annual surveillance audits. Often required in enterprise procurement and procurement frameworks worldwide.

First international management system standard dedicated to artificial intelligence, published by ISO/IEC in December 2023. Specifies requirements for an AI Management System (AIMS) covering policies, organization, leadership, planning, support, operation, performance evaluation, and improvement. Designed to be aligned with EU AI Act and NIST AI RMF. Certification just becoming available in 2025; expected to gain traction as a procurement signal alongside ISO 27001.

Global baseline sustainability standards issued by the International Sustainability Standards Board (ISSB) in June 2023: IFRS S1 (general sustainability-related disclosures) and IFRS S2 (climate-related disclosures). Adopted or being adopted by 30+ jurisdictions worldwide (UK, Canada, Australia, Japan, Brazil, Singapore, Hong Kong, Saudi Arabia). Designed to be interoperable with EU CSRD/ESRS and aligned with TCFD recommendations.

Generic English term for companies whose core business is selling IT services (consulting, implementation, application management, managed services). Encompasses systems integrators, application services providers, and digital agencies. Examples: Accenture, Cognizant, Infosys, Capgemini, CGI, TCS, Wipro, EPAM, plus thousands of mid-sized regional firms.

ServiceNow product suite covering infrastructure visibility (Discovery), service mapping, event management, AIOps, cloud governance, and operational intelligence. Complements ITSM by providing the operational data backbone (CMDB, dependencies, performance) needed to run IT services proactively. Often combined with monitoring tools (Datadog, Dynatrace, New Relic).

Discipline structuring how IT services are designed, delivered, operated, and improved, often aligned with the ITIL framework. Core processes: incident management, problem management, change management, request fulfillment, configuration management, knowledge management. ServiceNow is the dominant enterprise platform; alternatives include Atlassian Jira Service Management, BMC Helix, Cherwell.

Fanuc's high-level programming language, complementing the lower-level Teach Pendant Programming (TPP). Resembles Pascal. Used for complex motion planning, error handling, and sensor integration on Fanuc R-30 controllers. Fanuc dominates global automotive welding and assembly markets. ROBOGUIDE is Fanuc's offline programming and simulation environment.

ATLAS principle: mandatory validation checkpoints between project stages where a no-go decision halts the project rather than letting it drift. Each kill gate has explicit criteria (functional parity verified, discrepancies log signed off, performance baseline met, security review passed). Prevents the classic legacy migration disease: a project planned at 12 months becoming a 36-month program with cost multiplied by 3.

Proprietary programming language for KUKA industrial robots, syntactically inspired by Pascal. Programs are split between .src (logic) and .dat (data) files. Runs on KUKA KR C controllers. Dominant in European automotive, aerospace, and metalworking. Supported by KUKA.OfficeLite and KUKA.Sim Pro for offline programming and simulation.

IEC 61131-3 programming languages for Siemens SIMATIC PLCs (S7-300, S7-400, S7-1500). Ladder Logic (LD) is graphical, derived from electrical relay diagrams; SCL (Structured Control Language) is text-based, similar to Pascal. Widely used to drive industrial automation and integrate robot cells. PLCSIM Advanced enables offline simulation.

Foundation model trained on massive text corpora to understand and generate natural language, typically with billions to trillions of parameters. Frontier models in 2026 include GPT-5 (OpenAI), Claude Opus 4 (Anthropic), Gemini 2 (Google), Llama 4 (Meta), Mistral Large. Capabilities cover reasoning, coding, summarization, translation, agentic task execution. Cost economics shifted dramatically downward 2023-2026 enabling enterprise deployments at scale.

Proposed standard introduced by Jeremy Howard (fast.ai cofounder) in September 2024, suggesting websites publish a structured Markdown file at /llms.txt summarizing their content for LLM consumption. Analogous to robots.txt but for AI models. Adopted by Anthropic, Mintlify, and a growing list of technical documentation sites. Companion file llms-full.txt holds the full content. Becoming a baseline GEO best practice.

Engagement model where the provider takes accountability for an outcome (SLA, scope, deliverables) on a fixed-price or fixed-monthly basis, rather than billing for time. The provider manages the team, the methodology, and the risk. Equivalent to French CDS (Centre de Services). Highest provider commitment; requires mature scope definition.

Open protocol introduced by Anthropic in November 2024 standardizing how AI assistants connect to data sources, tools, and applications. Solves the M×N problem of bespoke integrations by providing a common interface. Adopted by major IDEs (Cursor, Windsurf), AI assistants (Claude, Cline), and a growing ecosystem of MCP servers exposing systems like GitHub, Slack, Postgres, file systems. Becoming a baseline for enterprise AI tool integration.

System managing shop-floor operations between ERP and equipment: production scheduling, work order tracking, material genealogy, quality management, OEE collection, operator interfaces. Major vendors: Siemens Opcenter, Rockwell FactoryTalk, GE Proficy, Aveva, SAP MES. Frequently the integration backbone for AI-driven manufacturing initiatives.

Provider that takes ongoing operational responsibility for a defined IT scope (infrastructure, security, cloud, applications, end-user support) under a recurring contract with SLAs. Distinct from project-based IT services firms. Major MSP segments: managed security service providers (MSSPs), cloud MSPs, application MSPs. The MSP market grew with cloud adoption and is now consolidating.

Delivery model where teams operate from a nearby country with cost advantage, time-zone alignment, language proximity, and cultural compatibility. From France or the UK, Tunisia is a primary nearshore destination (UTC+1, French and English fluency, large engineering pool). Co-delivery means the nearshore team works hand-in-hand with the client's onshore team rather than operating in isolation.

European Union directive on cybersecurity adopted on 14 December 2022, replacing NIS 1 (2016). Significantly expands the scope: essential entities (energy, transport, banking, financial market infrastructure, health, water, digital infrastructure, public administration, space) and important entities (postal, waste, chemicals, food, manufacturing, digital providers, research). Tiered incident notification: 24 hours early warning, 72 hours notification, one month final report. Sanctions up to €10M or 2% of worldwide turnover for essential entities.

Voluntary framework published by the US National Institute of Standards and Technology in January 2023 (AI RMF 1.0) to help organizations design, develop, deploy, and use AI systems trustworthily. Structured around four functions: Govern, Map, Measure, Manage. Complemented by AI RMF Generative AI Profile (July 2024) addressing generative AI risks. Increasingly required by US federal procurement, state regulators, and major enterprises as a de facto standard.

Updated version of the NIST Cybersecurity Framework, released February 2024. Six functions: Govern (new), Identify, Protect, Detect, Respond, Recover. Significantly expanded from CSF 1.1 (2018). Used as a foundational reference by US federal agencies (per Executive Order 13800), critical infrastructure sectors, and increasingly by international organizations. Often paired with ISO 27001, SOC 2, and sector-specific frameworks.

New York City local law in effect since 5 July 2023 regulating Automated Employment Decision Tools (AEDT). Requires employers and employment agencies using AEDTs to have an annual independent bias audit, publish summary results publicly, and notify candidates and employees of the AEDT use. The first US municipal-level AI bias audit law to enter active enforcement.

Manufacturing performance metric combining three dimensions: Availability (planned production time minus stops), Performance (actual cycle time vs ideal), and Quality (good output vs total output). World-class OEE is typically considered around 85% in discrete manufacturing. AI-driven optimization targets OEE leaks (micro-stops, slow runs, quality losses) that are invisible in conventional reporting.

System orchestrating the order lifecycle end-to-end across channels: order capture, validation, allocation across stocks and warehouses, fulfillment routing, returns. Critical for omnichannel retail and complex e-commerce. Major vendors: Manhattan Active OM, IBM Sterling OMS, Salesforce Order Management, Oracle Retail. AI augments OMS with smarter allocation and fraud prevention.

Canadian federal privacy law applying to private-sector organizations collecting, using, or disclosing personal information in the course of commercial activities. In force since 2001. Enforced by the Office of the Privacy Commissioner of Canada (OPC). Quebec, Alberta, and British Columbia have substantially similar provincial laws. The proposed Consumer Privacy Protection Act (CPPA, Bill C-27) would replace PIPEDA, but its adoption has been pending.

Practice of designing, structuring, and refining prompts to elicit desired behavior from LLMs. Includes techniques like few-shot examples, chain-of-thought, role assignment, structured output formats (JSON), system prompts, and prompt caching. Increasingly evolving toward context engineering: programmatic assembly of context windows from RAG, memory, tool outputs, and user input. Skills moving from individual practitioner art to systematic engineering discipline.

Quebec privacy law adopted September 2021, with phased entry into force 2022-2024. Substantially aligns Quebec with GDPR-style obligations: consent, data subject rights including a right to data portability (since September 2024), mandatory privacy impact assessments for high-risk processing, breach notification, automated decision-making transparency, and a privacy officer designation. Applies to all organizations operating in Quebec, including non-Canadian entities.

AI architecture combining a vector knowledge base with a language model. At query time, the system retrieves relevant passages from the knowledge base and injects them into the prompt sent to the LLM, which generates a response grounded in retrieved documents. Reduces hallucinations, enables source citation, keeps knowledge fresh without retraining. De facto standard for enterprise AI applications: customer support, knowledge management, legal research, internal Q&A.

ABB's proprietary programming language for IRC5 and OmniCore controllers, inspired by Pascal and ADA. Strong typing, structured tasking, multitasking support. Programs use .mod (modules) and .sys (system) files. RobotStudio is ABB's offline programming and simulation environment, with a free education edition. ABB has a major footprint in automotive welding, painting, and material handling.

Saudi Arabia's first comprehensive data protection law, fully effective since 14 September 2024 after a transition period. Enforced by the Saudi Data and Artificial Intelligence Authority (SDAIA). Establishes consent, transparency, data subject rights, and cross-border transfer requirements (transfers require adequacy decision or safeguards). Sanctions up to SAR 5 million per violation, with criminal penalties for sensitive data breaches. Often combined with KSA NCA cybersecurity controls (ECC) for full compliance.

Vocabulary of structured data tags created jointly by Google, Bing, Yahoo, and Yandex in 2011, used to mark up web pages so search engines and AI systems can understand content. Implemented as JSON-LD (recommended by Google), microdata, or RDFa. Covers types like Article, Product, Organization, Person, FAQPage, Event, Recipe, DefinedTerm. Foundational layer for both classical SEO rich results and modern Generative Engine Optimization.

Rules adopted by the US Securities and Exchange Commission in March 2024 requiring registrants to disclose material climate-related risks, governance, strategy, and risk management, with phased greenhouse gas emissions disclosures (Scope 1 and 2 for accelerated and large accelerated filers). The rules have been subject to ongoing legal challenges and stayed pending litigation, with implementation timeline uncertain in 2025-2026. Often analyzed alongside California SB 253 / SB 261 and ISSB standards.

ServiceNow ITOM module that builds and maintains visual service models linking business services to their underlying technical components (apps, servers, databases, networks). Foundational for impact analysis (which business service is affected by an incident?), proactive change management, and outage prevention. Top-down (business service first) or pattern-based discovery.

ATLAS deployment pattern: legacy and modernized systems run in parallel on the same inputs. Outputs are compared automatically; divergences trigger entries in the discrepancies log. The modernized system carries no production effect during shadow mode. Once divergences fall below an agreed threshold and stabilize, switchover happens with high confidence and a tested rollback path.

Audit framework developed by the AICPA (American Institute of Certified Public Accountants) for service organizations handling customer data. Based on five Trust Services Criteria: Security, Availability, Processing Integrity, Confidentiality, Privacy. Two report types: Type I (controls design at a point in time), Type II (controls effectiveness over a period, typically 6 to 12 months). De facto baseline for B2B SaaS vendors selling to US enterprise customers.

Engagement model where individual specialists are placed within a client team and managed by the client's project structures. The client retains delivery accountability; the provider supplies skills, day-rate billing, and continuity. Equivalent to French AT (Assistance Technique). Lowest contractual commitment from provider, highest flexibility for client.

Largest, globally recognized systems integrators capable of taking on prime contractor roles for complex multi-year programs (typically over USD 50 million). Examples: Accenture, Capgemini, Deloitte, IBM Consulting, CGI, Cognizant, TCS, Infosys. Often selected through mandatory procurement frameworks. Smaller firms partner with tier-1 integrators in co-delivery arrangements.

UK legislation enacted October 2023 imposing duties of care on online services hosting user-generated content or providing search functionality. Ofcom is the regulator. Three tiers based on size and risk: Category 1 (largest user-to-user services), Category 2A (largest search services), Category 2B (others above threshold). Phased enforcement throughout 2024-2025 covering illegal content, content harmful to children, and (controversially) age verification for adult content.

Universal Robots' programming language for collaborative robots (cobots) UR3, UR5, UR10, UR16, UR20, UR30. Python-like syntax, scriptable through the Polyscope teach pendant or external code. Backed by URSim, a free Linux-based simulator. Universal Robots is the leading cobot manufacturer worldwide, with strong adoption in SME manufacturing and packaging.

Software development practice where engineers work in continuous dialogue with a generative AI assistant (Claude Code, GitHub Copilot, Cursor, Windsurf, Replit Agent) to produce, refactor, test, and document code. Term coined by Andrej Karpathy in February 2025. Productivity gains of 2 to 3 times observed on legacy migration and documentation work, when paired with strict guardrails: tested suggestions, mandatory human review, enterprise-hosted models, and clear usage policy.

System managing warehouse operations: receiving, putaway, storage, picking, packing, shipping. Optimizes labor, slotting, space utilization. Modern WMS integrate robots and AGVs (autonomous guided vehicles). Major vendors: Manhattan Associates, Blue Yonder, SAP EWM, Oracle WMS Cloud, Reflex. AI applications include slotting optimization, demand-driven labor planning, and damage detection.