AI automation
The combination of AI capabilities with rules, systems and workflows to handle defined process steps.
AI becomes part of a real workflow rather than a single chat.
Glossary
AI terminology in plain business language.
Dependable definitions for executives, business teams and IT.
A shared vocabulary.
The glossary defines essential terms in the context of production business solutions.
AI agent
A digital system that can break an objective into steps and perform approved actions within defined boundaries.
An agent does more than answer; it works through an assignment.
AI employee
An organisational description of a specialised digital role with tasks, capabilities, knowledge and escalation.
A clearly bounded digital function, not a virtual person.
RAG
Retrieval-augmented generation connects a request with relevant passages from approved knowledge sources.
The AI receives enterprise knowledge selected for the question.
Multi-agent system
Several specialised agents handle separate parts of a shared process and hand outcomes over in a structured way.
Digital roles collaborate through a division of work.
Human in the loop
An operating model in which people review, approve, correct or stop at defined points.
AI works, but people intervene wherever accountability requires it.
Human oversight
Proportionate human supervision over the use, boundaries and outcomes of an AI system.
Accountable people retain meaningful control.
Agent memory
Bounded working or historical context an agent is allowed to use for an assignment.
The agent retains only what is intended for its task.
Tool use
Execution of explicitly approved functions or actions by an agent.
The agent may use specific tools, not act without limits.
API / interface
A defined connection through which applications exchange data or actions under control.
A governed connection path between software systems.
Orchestration
Coordination of process steps, digital roles, rules and handovers.
It determines who handles which task and when.
AI governance
The framework of roles, rules, controls and ownership for responsible AI use.
Who may do what, who reviews and who is accountable?
Data minimisation
Limiting processed information to what is necessary for the stated purpose.
Use only the data genuinely needed for the task.
Business process automation
End-to-end automation of business workflows across rules, systems, AI and approvals.
The coherent operating process is designed, not just one step.
Escalation
Orderly handover of a case to an accountable person when boundaries or criteria are reached.
The agent knows when to stop and ask for help.
Which process should work better in your organisation?
We begin with the work, the people and the current process — then assess which form of AI genuinely makes sense.