The language of enterprise AI
Definitions for the terms that come up when organizations move from AI tools to AI infrastructure — written plainly, without the vendor gloss.
Architecture
Enterprise AI intelligence layer
An enterprise AI intelligence layer is unified infrastructure connecting an organization's AI models, agents, workflows, and knowledge into one operating system. Unlike point-solution AI tools, it acts as connective tissue: routing tasks to the right model, enforcing governance centrally, and making organizational context available across every function.
Read the definitionOrganizational knowledge
Organizational knowledge is the accumulated context of how a company works: documents, decisions, policies, customer history, and undocumented convention. In an AI system it is the material that makes answers specific to your organization rather than generically correct, and it must be retrieved subject to permissions.
Read the definitionRetrieval-augmented generation (RAG)
Retrieval-augmented generation is a technique where a model retrieves relevant documents from a knowledge source and uses them to ground its response. Rather than relying only on what the model learned during training, it answers from material supplied at query time, which allows citation and keeps answers current.
Read the definitionVendor lock-in (AI)
Vendor lock-in in AI is dependence on one provider's models and interfaces to the point that switching requires rebuilding the systems above them. It matters more than in most software categories because model capability and pricing move quickly, so a fixed choice ages faster than the applications built on it.
Read the definitionAgents
Agentic enterprise
An agentic enterprise is an organization whose systems carry out work rather than only recording it. AI agents operate across business processes and applications, completing routine tasks, drafting outputs, and routing approvals, with people supervising outcomes instead of performing every step themselves.
Read the definitionAI agent
An AI agent is a system that pursues a goal across multiple steps, choosing actions and using tools rather than returning a single response. Where a model answers a question, an agent decides what to do next, acts on that decision, observes the result, and continues until the goal is met or it stops.
Read the definitionMulti-agent system
A multi-agent system coordinates several specialized AI agents on one objective, each handling part of the work and passing results between them. Coordination covers planning, delegation, and reconciliation, so the system completes tasks that cross functions rather than confining each agent to a single narrow job.
Read the definitionModels
Context window
A context window is the amount of text a model can consider at once, covering the instructions, retrieved documents, and conversation history supplied with a request. Anything outside it is invisible to the model, so what gets selected into the window determines the quality of the answer.
Read the definitionFine-tuning
Fine-tuning further trains an existing model on additional examples to adapt its behavior, tone, or format. It changes the model's weights, unlike retrieval, which supplies information at query time. Fine-tuning suits consistent behavior; retrieval suits current facts.
Read the definitionInference
Inference is a single run of a trained AI model on an input to produce an output. It is the unit of consumption in production AI: cost, latency, and governance are all measured per inference, which is why routing decisions and access controls are applied at this level rather than per application.
Read the definitionModel orchestration
Model orchestration is the routing of each AI task to the model best suited to it, rather than sending every request to a single general-purpose model. Routing decisions consider the nature of the task, cost, and latency, and happen automatically so users describe what they need instead of choosing a model.
Read the definitionGovernance
AI governance
AI governance is the set of controls determining who may use which AI capabilities, on what data, with what oversight, and with what record. Governance is effective only when enforced by infrastructure: a policy requiring review means nothing if the tooling does not make review mandatory and traceable.
Read the definitionHuman in the loop
Human in the loop describes an AI workflow where a person reviews or approves before an action takes effect. In practice it is applied proportionately: low-stakes work runs unattended, medium-stakes work receives lightweight approval, and irreversible or externally visible actions require structured review.
Read the definitionPrompt injection
Prompt injection is an attack where instructions hidden inside content a model reads cause it to behave as the attacker intends. Because a model cannot reliably separate instructions written by the user from text embedded in retrieved documents, any channel that feeds the index becomes a potential attack surface.
Read the definitionZero-trust AI
Zero-trust AI applies the principle of never trust, always verify to AI systems. Every inference request is authenticated regardless of origin, data access is authorized at the retrieval layer rather than the interface, outputs are filtered, and all activity is logged immutably so actions remain attributable.
Read the definitionOperations
AI observability
AI observability is the ability to see what an AI system did and why. It covers logging of who requested an inference, which model ran, what data was accessed, what was produced, and what action followed, in a record complete enough to reconstruct a decision after the fact.
Read the definitionAI sprawl
AI sprawl is the accumulation of disconnected AI tools across an organization, each with its own data access, governance, and cost. Every tool is individually justified, while collectively they produce fragmented context, inconsistent controls, and an audit surface that grows faster than the value delivered.
Read the definitionData residency
Data residency is the requirement that data be stored and processed in a specified geography. In AI systems it constrains where inference runs as well as where records sit, because sending a prompt to a model in another jurisdiction is itself a transfer of the data it contains.
Read the definitionOn-premise AI
On-premise AI runs models and supporting infrastructure inside an organization's own environment rather than a vendor's cloud. It is chosen where data sovereignty, regulation, or air-gapped operation prevents sending data to a third party, and it trades operational convenience for control.
Read the definition// Get Started
Ready to build smarter?
One secure platform for AI models, agents, and enterprise workflows.
Live in days, not months. No vendor lock-in.