Field Guide Topic
How humans and AI work together
This is the part of the vocabulary that changes fastest, and the part people most often use without agreeing on what it means. I think the useful questions are simple ones. Who decides? Who reviews? What context does the system actually have? What happens when the work is wrong?
The entries here move from the shape of the working relationship — human–AI collaboration and human-in-the-loop — into the mechanics of AI agents, agentic workflows, orchestration, and context engineering, and then out to the team and system level with AI-native teams and systems of intelligence. Read them together and you get a fairly practical picture of how this work gets done when it is done well.
Start with
Human–AI collaboration
Human–AI collaboration is a way of working in which people and AI systems contribute different capabilities to a shared goal. AI may retrieve information, identify patterns, generate options, or execute repeatable steps. People provide intent, context, judgment, creativity, empathy, and accountability. The value comes from the design of the partnership—not simply from giving a person access to an AI tool.
Read the entry →Human-in-the-loop
Human-in-the-loop, often shortened to HITL, is a way of designing an AI system so a person remains involved at moments where judgment, context, risk, exception handling, or accountability matter. The human does not need to approve every small action. The system should route the right decisions to people and allow routine, bounded work to continue without unnecessary friction.
Read the entry →AI agent
An AI agent is a software system that can pursue a goal by interpreting context, deciding what to do next, using tools, and taking actions with some degree of autonomy. Unlike a basic chatbot that primarily returns a response, an agent can work across multiple steps, observe results, and adjust its next action within defined boundaries.
Read the entry →Agentic workflow
An agentic workflow is a sequence of work in which one or more AI agents can decide how to pursue a goal, choose tools, take actions, evaluate intermediate results, and adapt their next steps within defined limits. It combines some AI autonomy with workflow structure, permissions, context, observability, and human checkpoints.
Read the entry →AI orchestration
AI orchestration is the coordination of models, agents, tools, data, rules, workflows, and people so they operate as one coherent system. It determines which capability should do what, in what order, with what context, under which permissions, and how work should be handed off, evaluated, or escalated.
Read the entry →Context engineering
Context engineering is the practice of giving an AI system the right information, instructions, tools, memory, examples, and constraints at the right moment so it can produce a useful result. It goes beyond writing a clever prompt by designing the complete information environment in which the model or agent operates.
Read the entry →AI-native team
An AI-native team is designed around the assumption that AI is an active part of how work is understood, assigned, performed, checked, and improved. It does not merely give existing roles access to copilots. It redesigns responsibilities, workflows, team shape, management practices, and measures of success around combined human and machine capability.
Read the entry →System of intelligence
A system of intelligence is a living layer that combines data, institutional knowledge, context, signals, reasoning, and feedback to help people and AI decide what to do next. Unlike a system of record, which primarily stores what happened, a system of intelligence interprets what is known and connects that understanding to coordinated action.
Read the entry →