A simple protocol for figuring out what actually works

You've probably tried at least one AI tool this year. Maybe it saved you twenty minutes on something. Maybe it produced something that looked useful until you actually read it. Maybe you spent an hour trying to get it to do something it clearly couldn't, closed the tab, and haven't opened it since.

What you almost certainly didn't do is compare notes with a colleague about any of that. And that's the problem.

Right now, most teachers are navigating AI tool adoption in complete isolation. Teachers are more likely to teach themselves how to use AI than to receive training from their schools, and only 19% say their school has any policy on AI use. In open-ended responses, many educators noted that they haven't been trained on the technology and haven't received guidance from district or school leaders. "Teachers shouldn't be left out in the wind on this issue," one high school social studies teacher wrote. And yet that's exactly where most of them are — solo, figuring it out, with no structure for sharing what they find (even now, almost two years later!).

The result is that every teacher is reinventing the same wheels, making the same discoveries, and abandoning the same tools for the same reasons — without any of that experience becoming useful to anyone else.

This post is about fixing that. Not with a committee, a task force, or a professional development session. With your colleagues, a shared document, and a text message.

– The Shift

The shift is small but it matters: from each person figuring this out alone to two or three people running parallel experiments on the same week, comparing what they actually found, and building a shared picture that's worth more than any of their individual experiences.

This doesn't require a meeting, a shared agenda, or an agreement about what AI is or isn't good for. It requires one shared document and the willingness to be honest about what happened when you tried something.

The norm from the start: nothing in the shared space is a recommendation or an endorsement. It's a log. "This was a waste of forty-five minutes and here's why" is exactly as valuable as "this saved me an hour." The point is honesty, not performance.

– The Framework

1. A shared document — one page, never longer

Three columns. That's it.

| The task I tried it on | What it produced | What I actually did with it |

Each person adds one row after trying a tool on something real. No stars, no ratings, no verdicts. Just: here's what happened when I tried this on this task.

The first column forces specificity. "Communication" is not a task. "Draft the weekly parent update email" is. Specificity is what makes the document useful.

The third column is the most important one. What you actually did with the output (revised it heavily, used it as-is, discarded it, used pieces of it as a starting point) tells you more about whether the tool is genuinely useful than any other metric. A tool that produces something requiring an hour of editing to become usable is not saving you time.

2. A check-in thread — one sentence per person

A group text for the moments right after you've tried something, when you have an immediate reaction that might benefit from a quick gut-check before it becomes a row in the document.

"Tried it on exit ticket comments. Feels like it's saving time but I'm not sure the suggestions are actually good — am I just accepting them because they look fine?"

"Used it for the parent email. Needed so much context before it got anywhere useful that I'm not sure what to log. Did you find that too?"

That second-person check — did you find that too? — is what the thread is for. It's the difference between a private frustration that disappears and an observation that gets pressure-tested by someone with relevant experience before it solidifies into a conclusion.
Sometimes what feels like a tool failure is actually a framing issue someone else already figured out. Sometimes what feels like a time-saver looks less convincing once you say it out loud to another person.

3. A five-minute conversation — once a week

In person, in the hallway or after a meeting. What surprised you all this week? Did anything in the document match your experience or contradict it? What are your next steps? Keep it simple and structured so it doesn’t add baggage!

Your Product

After two or three weeks of this, you'll have something that no amount of AI professional development can produce: a document of real-world data from people who know your context, tested against tasks you actually do, in a subject area that actually matches yours. It can also be the foundation of a professional conversation and approach that doesn’t currently exist in many schools.

– The Honest Part

This only works if people bring their actual experience, including the failures and the ambivalence. A document that contains only the things that worked is a marketing brochure, not a professional resource.

It also requires starting small. Two people is enough. Three is fine. More than that and the accountability disappears into a crowd.
And the structural reality: peer-tested knowledge built this way will tell you what works in your context. It cannot tell you whether a tool is safe, who has access to what data you put into it, or whether your district has any formal agreement with the company that made it. Those questions matter and they belong in a different conversation — the Learning Lab's post on AI and student data is a good place to start on those and this AI Privacy and Safety Checks by Sarah Wood is a great resource! This protocol is specifically about the professional workflow question: does this tool return more than it costs, in your room, for your tasks?

– Your Move

Identify one colleague. Not a committee. One person.

Send them this: "Want to try something? We each pick one task, try an AI tool on it this week, and add one row to a shared doc about what happened. Check in by text. Five minutes in person at the end of the week. That's it."

Then pick your task before you put your phone down. One task. One tool. One week.

If you tried this out at your school, did you make any updates to the shared doc that helped in sharing insights? What worked for you? Share in the comments!

---

This is part of Accingo's Collaboration Hub — practical frameworks for building teaching teams and partnerships.