Most school AI does not fail at the launch. It fails around day 90, when the excitement is gone, the pilot is still running, and nobody has decided whether it earned a place. This week's episode is about ending pilots on purpose. Plus three reads worth your time and a prompt library you can build in a week.
10x the context. Half the time.
Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.
Episode 4: The 90-Day AI Pilot
New episode this week, and it is the framework episode. I build my client work around a 30/60/90 AI Implementation Plan: Days 1 to 30 are Foundation, Days 31 to 60 are Cadence, Days 61 to 90 are Compound. The piece people skip is the very end, the day-90 checkpoint, and it is the most important part of the whole thing.
At day 90 you stop and make one of three calls about the workflow you piloted: scale it, keep it at the department level, or cut it. That is the whole decision. Not "let's keep going and see," which is how most pilots die. An actual choice, made out loud, with a reason. A pilot with no end date is not a pilot, it is a slow leak: it drains attention from the next thing and never gives you a clean signal about what is working.
The episode walks the whole 90 days, but if you take one thing from it: put a day-90 date on the calendar today and write down what "scale it" would have to look like. That one sentence will save you a year of drift.
The companion tools if you want to run this yourself: the 30/60/90 Fillable Workbook, the AI Pilot Plan Builder, and the Day-90 Keep/Kill/Scale Scorecard.
One more thing. This newsletter now has a room attached. I opened The Evalve Community: a free space where readers share what they are building with AI, ask implementation questions, and compare notes with people doing the same work. It even installs on your phone like an app, so the conversation travels. Come introduce yourself.
Security, AI Guidelines are Top School District Ed-Tech Concerns by EdWeek Market Brief. Security and AI guidance now sit at the top of the district ed-tech worry list, ahead of budgets in some responses. That matches what I hear on calls. The useful part is the gap it exposes: districts want guardrails but keep buying tools first. Read it before your next vendor meeting.
OECD Digital Education Outlook 2026 by OECD. The big international picture, and more readable than its page count suggests. The pattern worth stealing: systems that get value from AI invested in people and process before platforms. Skim the executive summary and the governance chapter. Skip the rest unless policy is your job.
Generative AI and IEP Goal Development: Implications for Special Education Teacher Preparation by CIDDL. A careful look at AI-drafted IEP goals from the teacher preparation angle. The takeaway I agree with: AI can draft, but the professional who knows the student has to drive. If your special education teams are already using AI for paperwork, and the data says most are, this is the framing to give them.
Learn AI in 5 minutes a day
You don't have to scroll every AI thread, track every new tool, or watch every demo.
The Rundown AI breaks it all down for you — the latest AI news, tools, and tutorials in one free 5-minute email every morning.
Trusted by 2M+ professionals at Apple, Google, and NASA.
Build a five-prompt starter library for one job.
A prompt library sounds like a big initiative. It is a shared document. Here is the one-week version, built for a single role: front office, case managers, instructional coaches, pick one.
Step one, collect. Ask three people in that role for the last patterned writing task they did: a parent email, a meeting recap, a translation request. You now have real work, not hypotheticals.
Step two, write five prompts. For each task, write a prompt with three parts: the role ("You are helping a school front office draft..."), the constraints (length, tone, reading level, what to never include), and a placeholder for the details. Test each one twice and fix what breaks.
Step three, put them where the work happens. A pinned doc, a shared drive folder, a tab in the tool your district approved. Not a new platform. If people need a login to reach the library, the library is already dead.
Step four, add a feedback line. At the bottom of the doc: "Used one of these? Tell me what it got wrong." That one line turns a static document into a system that improves.
The point is not the five prompts. The point is that a working prompt library tied to real jobs is the difference between staff using AI and staff using AI well. Start with one role, prove it, then let the next department copy the pattern.
If you want a running start, three of my nine-dollar prompt packs are prebuilt libraries for exactly these roles: Parent Communication, Meeting Prep and Summary, and SPED Documentation.
Did this issue land?
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