IDC - Definitions

IDC’s glossary of essential technology and market terms — each entry defines a concept clearly, backs it with IDC research, and links to the data behind it. Browse the full list below, or search for a specific term.
IDC’s AI assistant for querying a subscriber’s entitled IDC research and data in natural conversation, returning cited, source-traceable answers.
The five approved design pillars behind IDC Quanta — Embedded, Contextual, Secure, Aware, and Rigorous — each with one canonical, one-line definition.
A platform built from the ground up for conversational, structured-data access — not a chat interface added to a legacy research portal.
A workspace where a customer’s queries, documents, and outputs stay isolated and are never used to train an AI model.
The ability to trace an AI-generated answer back to a specific, verifiable source — including which document, version, and date it came from.
MCP is Anthropic’s open standard; IDC is building the first MCP server for the technology intelligence layer of the AI economy.
Every IDC Quanta answer passes through four specialized agents before it reaches you, including one that queries IDC’s canonical data directly.
Most AI tools answer questions. An agentic workflow executes the task end-to-end from a single prompt.
Entitlement transparency shows you exactly what your IDC subscription covers, and what it doesn’t, inside every IDC Quanta answer.
A common model and knowledge platform is the shared AI foundation that lets an organization train, tune and run multiple models in one governed environment.
Agent lineage and decision trace is the verifiable record of what an AI agent did, on whose authority, and which records it touched.
AI agent governance is the set of policies, controls and oversight mechanisms that keep autonomous AI agents operating safely within defined authority.
AI decision delegation is the practice of letting AI agents make defined business decisions within limits set in advance, with reversibility thresholds.
AI FinOps applies financial operations discipline to AI spend, including tokens, inference, agents and infrastructure, tagging every cost to an owner.
AI gaslighting is the risk that an enterprise’s own AI systems produce misleading or incorrect outputs at a volume humans cannot verify, eroding trust.
AI sovereignty is a nation’s or organization’s control over the entire AI ecosystem, including training data, models, algorithms and compute.
A knowledge architect defines the standards, practices and exception rules AI agents operate within, giving agents a governed foundation to learn from.
An AI-first enterprise runs a majority of its business processes with AI agents as the default starting point and humans assisting and providing oversight.
An AI-fueled organization is one where strategy, governance, people and technology are orchestrated so AI is embedded across the operating model.
A software factory is the set of systems that create, build, and deliver an organization’s code, increasingly treated as critical infrastructure.
Forward-deployed engineering is a delivery model in which provider-employed, customer-facing technical specialists work as an extension of client teams.
Human-on-the-loop AI supervision is a governance model in which people are notified of an AI agent’s actions and can intervene, while the agent acts.
Human-speed governance is oversight that depends on committees, scheduled reviews and manual reconciliation, a model that cannot keep pace with AI agents.
IDC MaturityScape for AI is a five-stage maturity model that shows how far an organization has progressed toward becoming AI-fueled.
Model routing sends each AI workload to the model that best fits its latency, accuracy and cost needs, so organizations can cut spend.
Physical AI shared responsibility is joint CIO, CTO and COO sponsorship of AI that controls robots and industrial equipment, so safety and downtime are owned together.
Policy-as-code governance writes governance rules as executable, machine-checkable code that runs automatically as data moves and agents act.
Techflation is the sustained, structural rise in the price of technology, led by memory, GPUs and networking, that forces IT leaders to budget more for AI.
Technical debt in AI initiatives is the accumulated legacy architecture, brittle integration and data debt that cause AI projects to fail or cost more.
AI thrivers and survivors are labels organizations give themselves for how they approach AI, and IDC’s scored maturity shows most self-described thrivers aren’t.
The hidden cost of AI-first is the revenue growth enterprises give up when they fund AI as a technology line item without matching investment in AI talent.
A wearable computing device is worn on the body, works hands-free, processes data digitally, and connects wirelessly to a network.
Agentic enterprise transformation professional services (AETP) are project-based advisory, design, and implementation services that help organizations plan, build, and scale agentic AI.
A dedicated cloud is an environment whose IT resources are dedicated to a single organization while still delivered with cloud attributes.
AI trust services are security services that treat AI systems as managed entities needing ongoing assurance across their life cycle, from model onboarding to runtime.
Universal connectivity is an enterprise network strategy that integrates multitransport redundancy, satellite connectivity, and edge inferencing for AI-driven workloads.
IDC classifies AI agents into four types by how they are built and deployed: in-app, standalone, custom-configured, and bespoke agents.
The decision complexity burden is the condition in which data volume and velocity exceed human capacity to process and act on it in time.
“AI is your audience” means answer engines and AI agents now read, compare, and summarize content before any person does, requiring structured content.