Welcome to the first issue of Evalve Out Louder, the newsletter that runs alongside the podcast. Each issue pairs the latest episode with a few things worth your time and one thing you can try this week. Short, useful, no fluff. Glad you're here.
Episode 1: Why Evalve Out Loud?
This first one is the why. Why I started Evalve, what the podcast is, and where I think AI in education is actually heading. The short version: only about 35% of teachers are engaged with AI right now, and we're already three years behind the curve. The people saying we don't need AI in schools are the same ones who, 30 years ago, said we didn't need the internet in schools. AI isn't going to take a teacher's job. But it can unload the parts of the work that drain your creativity and focus, so the adult in front of the student can do the part only an adult can do.
Teach students to argue with AI, not just ask it questions by Doan Winkel The trick isn't having students build a "Board of Advisors" inside ChatGPT or Claude. It's teaching them what to do next: pre-commit to a position, steelman the disagreement, journal what changed. Without that, the Board just becomes a fancier way for students to confirm what they already thought.
Canvas got hacked. The ransom isn't the story. by Dr. Aviva Legatt Instructure paid a ransom (reportedly around $10M) after ShinyHunters breached Canvas twice in nine days. Data exposed from 8,809 institutions and 275 million users. The part that should worry every education leader isn't the payment. It's how they got in: a free-tier signup pathway with weak verification, exploiting the same multi-tenant SaaS architecture every district's tools sit on. If you've never deliberately assessed your vendor concentration risk, now's the time.
Your best AI work is invisible by Nate Most schools have the same problem most companies do. Individuals get smarter with AI, the organization doesn't. The prompt that worked for one teacher dies in their chat history. The clever Claude project a counselor built never reaches the team next door. Shopify's fix is one design choice any district could copy: make non-sensitive AI work visible enough that the people around it can actually learn from it.
Stop asking ChatGPT or Claude to "write an IEP." That's too broad, and you'll get generic output every time. Instead, build the tool once. Open a Custom GPT or Claude Project and load it with the actual elements of strong IEPs at your district: the service language you use, your goal structure, your present levels format, the phrasing your team doesn't allow. Save that setup. Now every IEP draft starts from your standards, not a generic chatbot guessing what an IEP looks like. You spend the time you save on the parts only you can do.
The same move works for lesson plans, parent letters, MTSS notes, anything you write often where the format matters.
That's it for this one. If something here landed, share it with someone who'd get something out of it. If it didn't, that's part of the deal too. You can find more of what I'm working on at evalveconsulting.com, or book a call if you want to talk through what you're dealing with.
Talk soon, Chris