Goal
The outcome the agent is trying to reach — the thing that tells it when it’s actually done.
Think of it like
The destination in the sat-nav: every turn is judged by whether it gets you closer to it.
Example
The goal is “all tests pass and the feature works”; every edit and rerun is measured against whether that’s true yet.
How it actually works
The goal is the agent’s north star — it drives planning, shapes each decision, and defines the stop condition. A clear, checkable goal (“tests green”) makes agents work well; a fuzzy one (“make it better”) leaves them wandering or declaring victory early. Much of the trouble with agents traces back to a goal that was never crisply specified in the first place.
For product teams
A vague goal is the most common reason an agent “kind of” does the task — pin it down.
For engineers
The target outcome that defines the agent’s objective function and termination condition.
Related
- Subgoal — A smaller target on the way to the goal.
- Planning — Turning the goal into a sequence of steps.
- Task Decomposition — Breaking it down to make it actionable.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome