Guide / DELEGATE
From prompting to delegation: a simple AI agent brief
A short template for giving an AI agent a clear goal, the right inputs, safe limits, and a result a human can verify.
You commented DELEGATE, so here is the short workflow brief from the carousel.
The main idea is simple:
Give an AI agent a clear job, not a vague prompt.
The five-part brief
1. Goal
Describe the exact result you want. Name the final document, report, decision, or action.
2. Inputs
List the files, facts, examples, and systems the agent needs. Identify the source it should trust when information conflicts.
3. Process
Explain the essential steps and which tools the agent may use. Tell it what to do when information is missing.
4. Limits
State what the agent must never edit, delete, publish, purchase, or send. Require approval before sensitive actions.
5. Done
Define how a person will verify the result. Ask for evidence, links, or a short change summary when useful.
Copy this template
GOAL
Create:
For:
A good result looks like:
INPUTS
Use:
Trust first:
Ignore:
PROCESS
Follow these steps:
Allowed tools:
If information is missing:
LIMITS
Never:
Ask before:
DONE
Run these checks:
Return this evidence:
Final approver:
Start small
Choose one repeated workflow with clear rules and an output that is easy to review. A weekly report, lead-research brief, or data-cleanup task is a better first project than a large open-ended process.
Give the agent the minimum access it needs. Review every result until you understand its common failure patterns.
Further reading
The research describes Codex usage, not the entire workforce. Its human task-duration estimates are directional.
Published 9 August 2026.