B/D.

Lesson Nº2

ParentMap.

What level is your kid really at? Every homeschooling family I listened to ended up asking some version of that.

AI loop, no sound · 12s

  • Door: First, understand.
  • Research, product and direction of the build
  • Next.js + TypeScript
  • 2026
  • Live at parentmap.co

The blank page

What I kept hearing from parents who homeschool in Latin America is that nobody tells them how it's going. The parent is also the teacher, and there's no report card for that. A mom can finish a whole semester at the kitchen table and still go to bed wondering if she's doing it wrong, or if she's hurting her kids. I spent weeks reading what these families read and following the places where they ask for help before I touched the AI, and it always came back to the same practical question: what level is my kid really at? ParentMap started there, and in June I cut everything else until only the measurement was left.

The studio wall

scroll sideways →

  • ParentMap's front page asking ¿Tu hijo va al nivel que debería?, a no pressure assessment that shows how a child is doing against the level of their grade, subject by subject

    the question, on the product's front door

  • Sketchbook style page with a child's learning path drawn as a constellation of ink stars joined by an orange marker line, made with AI

    AI keyframe of the constellation, before animating

  • Deeper in parentmap.co: the section Ve el progreso a lo largo del tiempo, which invites parents to measure again in about three months and compare subject by subject

    the loop that brings parents back, live

Honest weaknesses

The crit

  1. The result comes from a single sitting. A kid having a bad morning can score like a kid who's behind, and I haven't yet built a way to tell those two cases apart.

  2. Since the product only measures now, it depends on parents coming back. The whole loop asks a parent to return in about three months and test again, and three months is a long time for a busy family to remember a website.

  3. I can't prove the measurement is accurate yet. So few families have used it that I have nothing to compare its results against. The analysis that will calibrate every question is already built, and it's waiting for enough real tests to have something to say.

Shipped & taught

  • Live at parentmap.co, in open beta
  • Measurement only since June 2026
  • Parents measure again in about three months
  • Built in Spanish first

Since late March 2026 the site has had almost 1,000 visits, about 980 sessions, without a single ad. Of those people, 26 came straight from an answer in ChatGPT or Copilot, and more than 100 started a placement test. Then one family paid, out of nowhere, and I had to go find out what it was.

What this taught me

The big decision on ParentMap happened before there was any code. These parents were scared, and I decided to answer that with evidence about their own kid, when most people in this space answer it by selling more material. I could only see that because I'd done the listening first. Back in June I wrote on this same page that the study plan was getting more attention than the diagnostic. A few days later I took my own crit seriously, hid the quizzes and the practice, and now the product only does the thing those parents were asking for from the start.

Next lesson · Nº3MUFU.