Real talk: I was tired of persuasive writing units that went nowhere. You know the drill. Students write generic essays. Maybe a few present to the class. Everyone forgets about it by Friday. Engagement hovers around 40% if you're lucky.
But this year? Twenty-eight of my 7th grade ELL students—from newcomers to advanced—built data-driven proposals and presented them to actual district nutritionists and Chef Gabriel. They answered tough questions. With confidence. In English.
How? I stopped searching for the perfect resource and started building exactly what my students needed.
The "Aha!" Moment
Here's what changed: Instead of asking Claude to "make me a persuasive writing game," I spent 30 minutes documenting my actual vision first.
In a Claude Project, I talked through:
- What my students needed to learn (persuasive writing + data interpretation + oral presentation)
- Who my students were (varying ELL levels, mostly Spanish-speaking and Hmong)
- What the real-world outcome looked like (presenting to district staff)
- What scaffolding they needed (bilingual support, sentence frames, academic vocabulary)
Claude helped me turn that conversation into markdown files saved in the Project's memory. This became my "source of truth."
Why this matters: Every tool I built after this pulled from the same vision. They weren't random games—they were a system that worked together.
The Three Tools (And Why Each One Mattered)
1. "Persuade the Chef" Presentation Guide
What it does: Walks students through building tri-fold boards step-by-step, with bilingual tooltips for academic vocabulary and a random question generator for practice.
Build time: 2 hours
The magic: Hover over "cost-effective" and get an explanation + example. Students actually started USING these terms in their presentations.
The bilingual toggle was clutch—newcomers could read instructions in Spanish but still produce work in English. And the random question generator? Students hit that button 50 times, practicing with partners until they felt ready.
2. Data Detective (Data Persuasion Challenge)
What it does: Competitive game where students see a chart, take a position, and present an oral argument. Teammates score them 1-4 points.
Build time: 3 hours
The magic: In 45 minutes, students got 5-7 reps making arguments out loud. More oral practice than most units give in a week.
Sentence starters (Claim → Evidence → Reasoning), academic vocabulary bank, 2-minute timer, peer scoring. Everything they needed for low-stakes practice before the high-stakes presentations.
3. Chef Gabriel's Inbox (Grammar Game)
What it does: Students review emails to Chef Gabriel for run-ons and fragments. Identify errors, choose the best fix, earn "Chef's Approval" bonuses.
Build time: 2-3 hours
The magic: Context. Every sentence was about cafeteria food—the same topic they were working on. Grammar practice didn't feel like a worksheet; it felt integrated.
Fifteen decisions per round. Instant feedback. A clear goal (11/15 correct). By the time students wrote their final proposals, run-ons and fragments had dropped significantly.
The Real Survey That Made It Real
Before building any tools, I surveyed the entire school. 236 students responded. We got actual data:
This data showed up everywhere—in practice games, in tooltips, in student proposals. Students weren't writing for me. They were advocating for their peers, backed by real evidence.
What Actually Happened
When district nutrition leaders walked in, students fielded questions like:
"What percentage requested this dish?"
"How does this reduce waste?"
"Is this cost-effective?"
And they answered. With data. In English. Including students who'd arrived that year as beginners.
The Progression:
- Chef's Inbox → Practice grammar in context
- Data Detective → Practice oral arguments with data
- Presentation Guide → Build the final product with confidence
How You Can Do This (Seriously)
Open Claude Projects. Talk for 30 minutes about your vision. Don't worry about being technical.
Ask Claude to create markdown docs covering your learning goals, student needs, assessment structure, and scaffolding requirements. Save these in Project memory.
Identify what you need that you can't provide 1-on-1 to 28 kids at once. Reference tools? Practice games? Immediate feedback?
Move to Claude Code. Say "Build X using the vision docs in project memory." Let it handle the technical stuff. You iterate on the pedagogy.
Deploy on Vercel (it's free). Get it in front of students. Watch what works. Iterate.
Total time: 10-15 hours over two weeks. Less time than hunting for resources that almost fit.
The Bottom Line
You don't need to be a coder. You need to know your students and be willing to try something new.
I'm not searching for the perfect resource anymore. I'm building what my students actually need. And now that I have the workflow, I can do it for every unit.
That's what AI should do for teachers. Not replace us. Not give us generic worksheets. But help us build what we've always wanted to build but never had time for.