6 min read
This is an illustrative story. Maya is a composite based on common situations solo music teachers describe, not a real person. The tool and its features are real.
It is 9:40 on a Tuesday night, and Maya Okafor is on her third text thread of the evening. One parent wants to know what their daughter should practice this week. Another says their son practiced “a lot” but cannot remember which pieces. A third asks, politely, whether the recital is still on the 14th. Maya teaches 26 piano students from a converted spare room, and the lessons themselves are the easy part. It is everything around them that eats her evenings.
She had tried the obvious fixes. A shared spreadsheet confused the parents. A paid studio management app did far more than she needed and cost more than she wanted. What she really wanted was small and specific: one page per student, showing this week’s assignment, a simple practice log the family could tick, and her lesson notes. Nothing else.
This is the story of how she built it herself in an afternoon, what went right, and the three things she would tell any solo teacher, tutor or coach thinking of doing the same.
The problem was never the lessons, it was the follow through
Maya added up a typical week. About 25 minutes a day went to answering “what should we practice?” messages, rewriting the same assignment notes by text, and chasing families before recitals. That is close to three hours a week of work nobody pays for, and it landed at the worst time of day.
Worse, the information lived in the wrong places. Assignments were scattered across texts. Her own notes sat in a paper notebook. When a student did well, nobody had an easy way to see the progress. When a student stalled, Maya often noticed late.
She had read our comparison of spreadsheets, free CRMs and AI assistants and recognized herself in it: the spreadsheet was free but nobody else could use it, and the bigger tools were built for studios ten times her size.
Building the app: one afternoon, one long description
On November 5, Cursor, a tool mostly known for helping programmers write code, announced that it can now build live apps from a description. You start what Cursor calls a cloud agent (an AI assistant that works on Cursor’s servers rather than your laptop), say you want to build an app, and describe it. Cursor sets up the web app, creates a database if one is needed, and arranges hosting. When you publish, you pick a unique web address and get a live link back, and individual plans can share an app publicly, which is how families can reach it without a Cursor account. The feature is available on Cursor’s paid plans with cloud agents turned on.
Maya is not a programmer. What she is good at is explaining things clearly, which is most of teaching. So she wrote her request the way she would write a lesson plan:
- Who uses it: me (the teacher) and each student’s family.
- What each family sees: only their own child’s page, through a private link I send them, with this week’s assignment, a practice log with a checkbox for each day, and my short note from the last lesson.
- What I see: a list of all students, with who logged practice this week and who did not.
- What it must never do: show one family another family’s information, or collect payments.
The first version took about 40 minutes to appear. It worked, but the practice log was buried at the bottom of the page. She asked for it at the top, asked for bigger buttons for phone screens, and asked for a “recital info” box she could update once for everyone. Each change took a few minutes.
The safety net she almost skipped
Here is the part of the story that matters most. Before anything goes live, Cursor runs automatic security checks, looking for problems such as passwords or secret keys accidentally left inside the app’s files. If a check fails, the agent explains what to fix. According to Cursor’s documentation, nothing is published until the checks pass and you choose to publish.
Maya’s first attempt did not pass. That was frustrating for about five minutes and reassuring after that. She also made two decisions of her own that no automatic check would have made for her:
- Collect as little as possible. Her students are children. The app stores first names and an initial only, no birthdays, no school names, no photos. If it ever leaked, there would be little to leak.
- Keep the old system for two weeks. She kept answering texts as usual while families tried the new page, so nobody was left confused if something broke.
She also added a short line to her studio policy explaining that the practice page was built with an AI tool and what it stores. Our guide to telling clients you use AI has wording you can borrow for exactly this.
What changed after a month
In this illustrative scenario, the results are the kind a teacher in Maya’s position could realistically expect:
- Evening messages dropped from a daily habit to a few a week, mostly about scheduling rather than “what should we practice?”
- Roughly two of the three lost hours came back, because assignments were written once, in one place.
- Stalled students became visible early. A row of empty practice checkboxes is hard to miss, and it gave Maya a kind, specific reason to check in with a family.
- Recital week was calmer. One updated box replaced 26 separate reminder texts.
Not everything was perfect. Two families never used the page and still preferred texting, and Maya decided that was fine. The goal was fewer interruptions, not a perfect system.
Three lessons for any solo teacher, tutor or coach
Describe the smallest useful thing
Maya’s app works because it does three things, not thirty. AI app builders will happily build features you never asked for if your description is vague. Write down who uses it, what each person sees, and what it must never do. That last list matters as much as the first two.
Treat client data like it belongs to someone else, because it does
The automatic checks catch technical mistakes. They do not decide what you should collect in the first place. Store the minimum, and be ready to explain it to a parent in one sentence.
Know when a ready made tool is the better buy
If you need payments, invoicing, calendar syncing and reminders, a dedicated studio tool will be safer and less work than building your own. A custom app makes sense for the narrow, specific gap that off the shelf tools fill badly. For a similar story about using existing tools instead of building, see how a solo photographer handled holiday bookings.
Have you built a small tool for your own business with AI, or are you still deciding whether it is worth the risk? Tell us what you would build first in the comments.
