Agent Guardrails

Agent Evaluation

Methods and metrics for assessing the performance, reliability, and safety of autonomous AI agents.

Human-in-the-Loop

A system design where human oversight and approval are required at critical decision points in automated processes.

Agent Guardrailing

Safety constraints imposed on AI agents that limit their action space, prevent dangerous operations, enforce budgets, and require approval for irreversible decisions.

Standardised protocols and languages used for inter-agent communication in multi-agent systems.

An AI agent that maintains an internal model of its world and reasons about actions before executing them.

An AI agent that maintains an internal representation of the world to inform its decision-making process.

An agentic architecture where a planning module decomposes goals into ordered tasks and a separate executor carries them out, enabling complex multi-step problem solving.

A workflow pattern where multiple specialised agents are sequentially connected, with each agent's output feeding the next.

Reasoning and Acting — a framework where language model agents alternate between reasoning traces and action execution.

The sequential composition of multiple AI agents where each agent's output becomes the input for the next, creating automated pipelines for complex multi-stage processes.

The transfer of a task or conversation from one specialised AI agent to another based on skill requirements, escalation rules, or domain boundaries.