Reference
AI agents
An AI agent uses a model to choose steps and tools while working on a task. What matters is what it can access or change, how you check its work and when it must stop.
Explore AI agentsTopic directory · XnMark knowledge
AI concepts and applications
Choose AI around a defined task and a measurable result. Understand prompts, workflows and agents before deciding how much autonomy a system needs.
From understanding to action
- 01Question
- 02Evidence
- 03Context
- 04Decision
Choose a useful starting point
If the task is drafting or transforming content, start with prompt engineering and define the source facts and acceptance criteria. If the system needs to choose actions, read the agent explanation and compare agents with workflows. The appropriate level of autonomy depends on the task, tools and review requirements.
Frequently asked questions
How should we use this section?
Choose one bounded use case before expanding the system. Compare accuracy, failure handling and operating cost with a simpler process. Keep an owner for exceptions and review material actions before external publication. These resources help frame an operating decision rather than promise dependable results from a demonstration.
What if we need help applying the guidance?
Use the linked guides and service pages to understand the scope, then describe your business challenge. A definition or example explains an approach; it is not a diagnosis of your own implementation.
Comparison
AI agents versus workflows
Use a workflow when the required steps are known and repeatable. Consider an agent when choosing the steps is part of the task. Compare correct completion, control and cost before choosing the more flexible architecture.
Explore AI agents versus workflowsReference
Prompt engineering
Prompt engineering means giving an AI model a clear task, useful context and output requirements, then checking the result. Better instructions can help, but they do not replace evidence or review.
Explore Prompt engineeringOur method
Research and evidence standards
Our research standard is to state the question, retain the evidence, distinguish facts from interpretations and keep important limitations visible. Page creation should follow a useful reader task rather than a target number of URLs.
Explore Research and evidence standardsMake the next decision clearer
What is holding back your growth?
Tell us what you are trying to improve, which markets matter and where the evidence is unclear.