Weekly schedule messages that write themselves, and no more cold rooms

An organization with instructors across several locations wrote a weekly schedule email by hand, and the locations themselves were never informed in a structured way. Now both messages go out automatically from the existing schedule.

The starting point

This organization’s instructors work at different locations and with changing groups. Every week a staff member wrote an email by hand: pulling each instructor’s own sessions out of the big schedule and listing them. About half an hour a week, all year round.

A second flow did not exist. The schools and venues used for rehearsals never got structured notice of when which room was needed. That ran through loose messages, with misunderstandings, last-minute changes and now and then an instructor in a cold or locked room.

What we built

A generator that reads the existing schedule file and turns it into two kinds of messages:

a personal weekly email per instructor, with only their own sessions: date, time, group and location. Anyone with nothing scheduled that week gets no email;
a message per location, with the days, times and rooms, each on the rhythm that location needs.

Plus an address list the organization manages itself, a log of what was sent, and an emergency stop to hold back a send.

The idea behind it

We changed nothing about how the schedule is made. The team keeps working in the file it knows, and the system only reads from it. That made adoption light: no new software to learn, no double entry.

Here too, no AI runs in production. Writing a weekly email is filtering, grouping and putting things in a table. Plain logic does that more reliably and more cheaply than a language model.

We built trust step by step. For four weeks the system only produced drafts, which were compared with the real emails. Only after written sign-off by the organization did the messages start going out automatically.

Expected results

The weekly half hour becomes a few minutes of review, roughly 22 net hours a year. That estimate already accounts for the new work, such as keeping the address list up to date. The first automatic send went out in mid-October 2026, to the locations at eight and to the instructors at nine.

Expected impact

The real gain is not the half hour. It is that the locations know, for the first time, correctly and on time, when to open and heat a room, and that instructors no longer have to puzzle their week out of a big schedule. Errors in the schedule also surface faster, because everyone involved now sees exactly what is planned for them.