After the Last Light

1,870 teachers stay up after their families fall asleep to coach themselves. Nobody asked them to.

Rumi. The 13th-century Sufi poet who taught millions to let go of their egos, stay grounded, and never stop learning. Eight hundred years later, his message hasn’t changed: the pursuit of knowledge is the pursuit of being fully alive.

That’s the spirit we wanted when we named our coaching tool. But let’s talk about what Rumi would actually see if he walked into a classroom today.

Walk into a government school in Pakistan and you’ll usually find one of three scenes playing out: Scene 1: Forty students singing “Assalamualaikum!” in perfect, robotic unison. Scene 2: The teacher writing word meanings on the board while students copy them into notebooks. Nobody is talking. Scene 3: Students reading aloud in the same flat monotone, as if reciting a shopping list instead of a poem.

None of these are teaching failures. They’re coaching failures. Because nobody ever showed these teachers what else is possible.

The Coaching Gap

Think about every profession that actually gets better over time. Medical residents spend four years under supervision. Every patient interaction reviewed. Pilots debrief after every flight. Now think about teaching. A government supervisor visits once a year. Fills out a form. Leaves. Nothing changes until the next workshop, which is really just another room where someone talks at you for six hours and sends you back to Monday morning alone.

The world spends billions on teacher training. We tested whether any of it changed what happens in classrooms. 306 observations across 138 schools. Of the 47 teachers observed three or more times, none showed sustained improvement in student talk time. Observation without coaching is a dead end.

93% of teachers in low-income countries never receive instructional coaching. Not because nobody cares. Because it’s impossibly expensive. Until now.

So We Built Rumi

A teacher finishes a lesson. Opens WhatsApp. Sends an audio recording of their classroom. Rumi transcribes it, analyzes it against evidence-based teaching frameworks, and starts a coaching conversation. Not a report. Not a score. A conversation. It asks the teacher to reflect. Identifies what worked. Suggests one thing to try tomorrow.

No app to download. No training required. Works on a basic phone over 2G. If it doesn’t work on the phone a teacher already has, it doesn’t work.

41% of teachers code-switch between Urdu and English. That’s how you know it’s real.

Two Tools. One Lesson.

We have two tools built on the same AI. Rumi is a WhatsApp coach teachers use on themselves. The Digital Coach is the same AI, but driven by a school supervisor who records the classroom observation. Same analysis, same coaching frameworks, but it depends on someone else being in the room. The Digital Coach produces fewer sessions. Not because it’s worse, but because one supervisor can’t reach hundreds of teachers. Remove the bottleneck, let teachers coach themselves, and the numbers speak for themselves.

Both tools run the same AI analysis. The Digital Coach needs a supervisor present. Rumi just needs a teacher and a phone. That’s the difference.

The recording happens during school hours. The coaching happens at night. 10pm is the single busiest hour. 37% of all audio sessions land between 6pm and midnight. Another 12% between midnight and 4am. Nearly 700 sessions in the dead of night, no one watching, no one keeping score.

In Sri Lanka, the pattern is different: 56% of chat sessions happen during school hours, with 11am alone accounting for 22% of all activity. Two countries, two adoption patterns, both voluntary.

The Cost Line

A single human coaching observation costs $40. That covers coach salary, transport, per diem, supervision, training. Rumi does it for $0.35. That’s 114 times cheaper. And the human coach maxes out at 40 observations a month during the 6 months schools are actually in session. Rumi never stops.

What the Data Actually Shows

We score every classroom observation across five dimensions of teaching quality. Here’s what 475 observations reveal:

Teachers are following the lesson plan (87%) and using good practices (90%). But students only speak for 9% of the lesson. The shape of this radar is the shape of the problem: teacher behavior is changing, student behavior hasn’t caught up yet.

Where Rumi Is Now

Last month we wrote about the forgotten layer. This month: the gap between lesson plans and classroom practice. Rumi isn’t the final answer. But when teachers voluntarily coach themselves at 10pm on a weeknight, and 37 new ones show up every day without being asked. That’s a signal worth paying attention to.

Vote: What Should We Write Next?

Each click sends a quick email. Takes 2 seconds. Tells us what you care about.

Taleemabad enables teachers to teach well so students learn.

www.taleemabad.com