AI SYSTEMS / AI AGENTS
An AI agent handles one recurring task and passes everything else to a person.
It suits work too variable for fixed rules but narrow enough to review: research, sorting incoming messages or documents, drafting replies, pulling data out of files. Each agent works within limits written before the build.
WHAT THIS TOPIC COVERS
A narrow job and a person who reviews it.
The topic covers the practical side of agents: which tasks suit one, where it sits next to ordinary automation, when it stops and asks a person, and what it takes to keep one running.
- Hand-over cases tested on real examples before launch
- A named person who reviews what the agent produces
- Running cost and output quality checked as part of the system
- Access to your tools through their API, never a person's login
QUESTIONS THIS PAGE ANSWERS
Questions about AI agents.
What is an AI agent?
Software that uses a language model to read variable input, choose among a limited set of actions and hand anything outside its scope to a person.
How is an agent different from a workflow?
A workflow repeats the same steps and gives the same result for the same input. An agent reads each input and picks its next action within limits agreed in advance. When only one step of a process needs judgement, the usual design keeps the workflow and places an agent at that step.
How do you scope an AI agent?
By writing down what it receives, what it produces, which actions it may take, when it must stop and ask, and who reviews its work. If the team cannot describe the job in two sentences, the scope is still too wide.
What does it cost to run an AI agent?
Model usage is only part of it. The tools it connects to, the time someone spends reviewing its output and the upkeep when models or connected tools change all add to the cost, so the volume of the task matters as much as the model chosen.
GUIDES
Guides on AI agents.
From choosing the task to deciding when several agents are worth it.
Multi-Agent Systems for Business
Practical guide to multi-agent AI systems for business: four architecture patterns, worked examples, failure modes and when several agents are worth it.
How to Scope an AI Agent
Framework for evaluating which business tasks are suitable for AI agents and which require human judgment.
RELATED TOPICS
Topics next to AI agents.
NEXT STEP
Have a recurring task in mind?
Each agent is quoted on the monthly hours it takes off your team, paid before the build and used under a licence; your accounts and data stay in your name. Describe the task, how often it runs and who reviews it today.
Talk to us