Last Week with Hypandra: From One Session to Three
A week ago we ran our first Curiosity Builds session with a teacher and shipped CuriousFractions.com. This week, two more instructors from the same school sat down with us and we shipped two more.
Two More Builds with Instructors
FindTheSlope.com came out of a session with a math teacher. She wanted a way for students to explore what slope actually means—not just compute it. We built interactive slope explorations, a lesson on coefficients of different functions, a lesson editor so she can create and preview her own content, a teacher invite system, and a review workflow. She can now assign, review, and iterate on lessons without touching code.
There are plenty of tools out there for slope graphs, math visualizations, and more. Under the hood we used the classic React Flow + Dagre combo: React Flow renders the interactive canvas, Dagre computes the node positions. What came out of this conversation was bringing them together into a repeatable setup that lets the teacher customize, scaffold, and review on her own.
HighSchoolResumes.com came from a high school science teacher who also helps students with college and career preparation. The challenge: most resume builders assume you already know what to put on a resume, and they aren't optimized to help you find ways to address the gaps. We built an interview-driven flow that asks students about their experiences and helps them articulate what they've done. It includes PDF export and a demo mode so the teacher can walk students through it. The centerpiece: an opportunity finder that surfaces curated and AI-facilitated live web search results: internships, volunteer positions, and extracurriculars matched to each student's interests.
Check out the AI Transparency section below to see how we made every AI decision in this build visible and teachable.
Three builds with instructors in two weeks, all from the same school. Each is a Curiosity Build—a hands-on intro to AI literacy and Hypandra's AI principles through brainstorming, planning, and building together with AI—with each session more ambitious than the last. It's what we meant when we wrote about demanding curiosity—building with others matters, and each session proves it.
Modeling AI Transparency
HighSchoolResumes.com gave us a chance to model something we've been thinking about: what does AI transparency look like in a tool that students actually use? What happens when a student can trace every AI suggestion back to its source?
We added clickable sparkle icons throughout the interface wherever AI is involved. Click one and you see the full prompt—exactly what was sent to the model and why. Every AI-generated suggestion carries provenance tracking so users can see what the model contributed. And we built a full AI transparency page that breaks down the models used, what data flows where, and what the AI can and can't do.
Alongside our changelog entries, we're trying to make AI use visible and teachable—especially for learners who are still forming their understanding of what these tools are and what they're doing.
WildReader's Engine Keeps Growing
WildReader is a learning toolkit for ages 3–12+ that Daniel originally developed for his three- and five-year-old. It shipped with literacy games—letter recognition, phonics, early decoding—and this week it crossed into math. New activities include dot arrays, ten frames, and a magnitude explorer where kids tap a number and see it laid out as scattered dots, rows, or wrapped grids. There's also a rate game where kids race against their own speed.
We opened up anonymous play: anyone can try games without signing up, with progress tracked locally. And the Card Maker—which started as a Valentine card creator—got a Lunar New Year theme and a basic mode for simpler designs.
Updates Across the Board
The older Curiosity Builds kept moving too.
Added custom word lists. Learners, teachers, and parents can now upload their own spelling words, and the system enriches them with reading levels and phonetic patterns. Study sessions pull from those custom lists, and a new "List" badge makes it clear when you're studying your own words versus the built-in library.
Expanded to 72 historical personas with new generated portraits. A new browse page with filters makes it easier to explore by era and region. And thanks to a user tip, we caught a bug in the timeline's mobile layout—text was scrunched and overlapping on smaller screens, now it reads cleanly.
SellSequence
Trevan: Last week, I got so caught up doing a Curiosity Builds demo that it turned into a much deeper project. While filming my idea for a scheduling app, I got inspired to make it so much more. I had five hours on a plane, and the plane wifi didn't disappoint.
Originally, I just needed something to help me sell my house. I'm in the process of doing it, and it's not something people do often. There didn't seem to be any good tools that let me get everything in one place: the scheduling, the sequencing, knowing what I need to do and when. Something that let me keep my real estate agent and my co-owner updated. And so SellSequence was born.
The app is a lightweight, easy-to-understand project management tool that interviews you or your agent, takes that interview, and builds a plan you can then customize. There's also a budgeting feature, and I keep adding things all the time. It's really exciting to apply my new skills, show what I'm doing, and build something that might actually be a viable product or business.
Consulting
We're launching a consulting product and service. The idea: start with a conversation, move into an audit of your current AI use—where it's adding value AND where your values might be at stake—and then build together. Not a report that dies in a PDF, but a working system your team can run, learn with, and keep iterating on. We're offering a free 45-minute conversation to start. More details on the page.
What's Next
The consulting page goes live. We have more Curiosity Builds sessions on the calendar. More builds, more feedback, more iteration—that's the rhythm now.
This post was written with the assistance of Claude Code (Opus 4.6) for editing and organization. All claims have been verified and the author takes full responsibility for the content. See hypandra.com/writing for our approach to writing with AI.