Running with our own teams first
The training finished.The problem didn't.
It was never about discipline — nothing you were taught was built around your actual work. This is built the other way round.
You finished it.
Certificate, full marks, and on Monday you still could not start the actual task.
Nobody noticed.
You went quiet in week two. The system still had you down as enrolled.
You asked the chatbot.
It answered fluently about a course it had never read, and you had no way to tell.
Week two
Nobody quits at the hard part.
You quit somewhere in the first third, on an ordinary Tuesday, and it never feels like quitting. It feels like being busy. You will come back to it at the weekend.
A better final exam would have helped nobody. What you needed was someone noticing in week two.
Learners still active, by module
- Reached the wall
- Never got this far — unreached, not harder
01Almost everyone starts.
So someone is reading. Every week.
Where you are, what is holding you up, and one thing to do about it. Not a score.
Your starting point is kept
Each week is recorded, not overwritten. Six weeks in, "how is this going" has six real answers instead of one number with the history behind it gone.
A named next step
Not "needs support." Sit with someone, pair up, or go back and practise one specific thing — something you can actually do on Monday.
A person decides it
The read is compiled automatically. A tutor decides what the next step actually is before anyone contacts you.
01Each learner in the cohort: progress, read row by row.
An aside
We tried taking the structure out.
Give people a blank chat window and total freedom and a few of them fly. We watched it happen. Most people go quiet within the week — not because they lack curiosity, but because nothing is asking anything of them.
Structure without relevance is school. Relevance without structure is a tab you closed. Neither works alone, and pretending otherwise is how most of this gets built. What is left is the harder version: enough structure to hold you, enough relevance to be worth holding.
Ustaad
A tutor that has done the reading.
It is eleven at night and you are stuck on one line. A general chatbot will not fail loudly here — it answers fluently about a course it has never seen, and you have no way to tell. Ustaad answers only from the material it was given, and says so when that runs out.
Sounds right. Traces back to nothing.
Same answer — and you can follow it back to what it was grounded in.
01Two answers. Both fluent, both confident.
Ask it anywhere
It has read the actual course
The module content itself, not a summary — so an answer can point at where it came from.
It says when it doesn’t know
Ask outside the material and it tells you, rather than improvising something fluent.
It keeps up with you
A term it explained early on gets used freely later, rather than re-explained.
What comes out the other end
You don't submit an essay. You ship something that runs.
You build for the job you already have — something your own week actually needs. A working tool aimed at real work is a harder thing to fake than a score.
Turns a week of scattered updates into the summary you would otherwise write by hand.
A drafting tool that knows the house style, so the first version needs editing, not rewriting.
Reads a pile of user feedback and comes out with what is actually worth arguing about.
Reads a sheet the way a reviewer would and says where the numbers stop making sense.
A lesson-prep companion that has read the curriculum it is meant to be helping with.
Keeps track of who is stuck on what, so the next conversation starts in the right place.
The kind of thing that gets built, not a gallery — the real ones go up once their authors have said yes.
If you're the one teaching
And if you're the one who has to teach it.
Right now that means links in WhatsApp, a folder nobody opens, and a weekend spent making slides — with no way to tell afterwards whether any of it landed. A course here takes weeks, not quarters. You can put one together for your team and still be the person who signs off on every word of it.
We use AI to build it, and would rather say so than be caught not saying it. What makes that safe rather than reckless is that the person who knows the subject has to approve it — and can send it back.
01Three stages draft the material in parallel.
A module has to make you do something, not just tell you something.
Assessment asks you to apply the idea — recall is easy to write and useless to measure.
Nothing generated goes live until a person who knows the subject has signed it off — and they can send it back.
Before you ask
Questions we get.
Can anyone sign up?
No. Access is by invitation and an administrator reviews every request. Tell us what you work on — that is the part that gets read.
How do I sign in once I have access?
Taleemabad and partner addresses sign in with Google. Everyone else enters their email and receives a one-time code by email or WhatsApp.
Who is this actually for?
Anyone with a job who needs to be able to do something new soon. It is built around your own work — with no work to apply it to, it will not land.
What stops someone just using AI to do the assignment?
Nothing, and we would rather build for that than pretend otherwise. What you hand in is a working thing aimed at your own job, and it gets read by someone who knows that job. Using AI to build it is the skill. Not being able to explain it is the tell.
Tell us what you're trying to get good at.
If you have read this far and something in it felt familiar, that is most of the qualification. A request is a short note about your work and what you want to be able to do next. An administrator reads every one.