Automating before they know what they’re automating. They spin up an agent without clear rules, expect it to “figure it out,” then blame the AI when it makes mistakes.
The Clarity Problem
It’s like hiring someone to do your job but never telling them how you do it. You’d never hire a coach without explaining your coaching philosophy, right? But tons of educators set up an AI agent to “answer customer questions” without actually writing down the answers, tone, or values they care about. The agent improvises. It goes wrong. And the educator says “AI isn’t ready for this.”
The mistake isn’t the AI. It’s launching before you’ve done the actual work: defining rules, writing examples, clarifying boundaries. You need to know what “good” looks like before you can ask an agent to produce it.
What Clear Setup Looks Like
Before you turn on an agent to answer customer questions, write down: the five most common questions you get and how you’d answer each. Your tone. Which questions you want to auto-respond to and which ones should escalate to you. Then show the agent those examples. “Here’s what good looks like.” The agent learns the pattern and replicates it.
This takes two hours, not two months. But it’s the difference between “this works” and “this is broken.” You see this with email sequences too. Educators set up a welcome automation without actually writing a compelling welcome email. Then they wonder why students don’t engage. The tool isn’t the problem. The input was.
What This Means for Educators
Before you deploy an agent, do the grunt work. Write the rules. Define the examples. Be specific. This is actually faster than troubleshooting a broken agent later. Most educators skip this step because it feels tedious. It’s not. It’s the thing that makes agents work.
The Clarity Rule
Know what you want to automate before you automate it. Write the rules. Show examples. Then turn on the agent. This separates success from failure.