Free Resource — Educators & Nonprofits

Agent
Accountability
Checklist

Five plain-language checkpoints for anyone using AI agents in work that serves students, clients, or communities. No technical background required.

📋

Download the checklist

One page. Print it, share it, use it before any agent runs in your organisation.

Download Free PDF

PDF · One page · Print-ready · No sign-up required

The five checkpoints

Checkpoint 01

Entry — Before you set an agent going

Is the task clearly defined?

Is the audience appropriate?

Have you chosen the right tool for this task?

Checkpoint 02

Instruction — What you tell the agent to do

Is your instruction specific enough to be executed faithfully?

Have you named what the agent should NOT do?

Are you comfortable if this instruction is fulfilled exactly as written?

Checkpoint 03

Override — Your right to stop, redirect, or reject

Do you know how to stop the agent if something goes wrong?

Is there a human being responsible for the outcome?

Do the people affected know an agent is involved?

Checkpoint 04

Audit — Reviewing the output before it becomes an action

Have you read the output before it goes out under your name?

Does the output reflect your organisation's voice and values?

Have you documented that this output was reviewed by a human?

Checkpoint 05

Exit — After the agent has completed its task

Did the output match the intent?

Was anything produced that should not be shared, stored, or acted upon?

What did you learn from this run?

Want more resources like this?

Get updates when new tools and guides are published — plain language, no jargon.

You are on the list.

This checklist reflects the Human-to-Agent (H2A) protocol — the interface between human intent and agent execution. It is designed to be reviewed in under three minutes and adapted to your specific context. Built by Pertinent — practical AI tools for people who work.

AgentAccountabilityChecklist

Want more resources like this?

Agent
Accountability
Checklist