Students understand
the material betterProfessors understand
their students better
Agora is a short spoken review with Glo, an examiner built from your own course materials. You explain what you learned out loud, and you get an honest read on what’s solid and what isn’t, while there’s still time to do something about it.
Your professor sees how the class is doing. Your session stays yours.
In pilot in fall 2026 courses at Harvard and MIT.
Glo is ready
Live transcript
Written as it is spoken
Glo
On the problem set you found that the amplitude of the damped oscillator decays exponentially. Why exponentially, rather than linearly?
You
Because the damping force goes with velocity, so the energy you lose each cycle is proportional to the energy you still have…
Glo
Right. Your professor drew that envelope in Thursday’s lecture and said it was the part to remember. It only holds under one condition. Which one?
You
The damping has to be light? Otherwise it doesn’t really oscillate, it just creeps back to zero.
Glo
Good. Now tie that back to the resonance curve you derived in week two. What happens to the peak as the damping grows?
Built from your course’s own materials
Reading it again is not the same as knowing it
You went to the lecture. You read the chapter. You got through the problem set, with a bit of help. Everything felt fine. And then the exam asks you to explain the thing in your own words and you discover, in the worst possible room, that you couldn’t.
Recognizing an idea on a page is easy and feels like understanding. Producing it out loud, unprompted, is the part that’s hard, and it’s the only real test of whether you have it. Most courses ask you to do that exactly twice a term, both times for a grade.
Agora asks you every week, for nothing.
How a session goes
01
Open this week’s review
Your professor brings the course materials they already teach from. Agora turns the week into the ideas that matter, and you see what they are before you start.
02
Talk it through with Glo
Out loud, in your own words. Glo listens to your whole answer and follows up: pushing further when you’re solid, coming at it from another angle when you’re lost.
03
Find out what you actually know
Feedback on each idea, grounded in what you actually said, so you can see why. Then a clear next step worth your time.
04
Come back to what didn’t stick
Your progress builds across the term, so the thing you fumbled in week three is something you meet again well before the exam does.
Private by design
An honest answer about what you don’t understand is only safe to give if it stays private. So we built it that way first, and everything else around it. Your professor wants to know where the class is, not to read over your shoulder, and Agora is built so they can have the first without the second.
Your transcript is yours
Your professor can’t open your session, your transcript, or your per-concept scores. They stay private to you.
Your professor sees the class
What reaches them is the class as a whole: which ideas landed, which didn’t, and what’s worth reteaching. Your name comes up there only if you’re doing really strong work or could use a hand, and then with a specific reason, not a score or a ranking.
And that you showed up
Participation is what your professor gets about you individually: that you did the work your course assigned, and in some courses your preparation record. Not your transcript. Being lost inside a weekly review doesn’t count against you.
Why not just ask ChatGPT?
Because it will tell you the answer, and you will believe you knew it. General AI is built to resolve your confusion as quickly as possible. That feels like learning and mostly isn’t: the work that makes something stick is the work it just did for you. Glo is built the other way round, to find the place you’re shaky and stay there until you’ve said something true about it.
There is a second thing a general chatbot can’t do. Paste your slides into one and it knows the subject. It still doesn’t know your course: which slide your professor spent twenty minutes on, what the problem set actually asked, what got called “exam-relevant” on a Thursday afternoon.
Glo knows, because the professor teaching the course is the one who put it there.
What Glo works from
The course’s own materials
The syllabus, the lecture notes and slides, the readings, the problem sets, the practice exams your class is given. Not the internet’s general idea of the subject.
Your professor’s own instructions
Before each week’s review is published, your professor tells Glo what to emphasise, what to add and what to cut. Whatever they say becomes a rule Glo follows, above its own judgment.
What was actually taught
In some courses Glo will even know what your professor said in the lecture itself: the aside that mattered, the example they worked through, the thing they told you would be on the exam.
Closer to a teaching assistant who sat in your lectures than to a chatbot that has read your textbook.
The questions students ask us
Can my professor read what I said?
Your session stays private to you. No one in your course, your professor included, can open your transcript, your per-concept statuses or your progress over the term. Your professor sees how the class is doing, whether you took part, and some feedback reports with names removed. In some courses, your professor may also see questions students asked, in their own words but without names. How that works.
Does this affect my grade?
That is entirely your professor’s call. What Agora gives them about you is participation: that you did the work your course assigned, and in some courses your preparation record, plus an occasional note if you’re doing really strong work or could use a hand. It gives them no transcript and no per-concept score. The feedback in your report is written for you to act on, not for anyone to grade you with.
What if I blank, or say something wrong?
Then you’ve found the thing worth studying, which is the entire point of doing this a week before the exam instead of during it. Glo follows up when an answer is thin and comes at an idea from a new angle when it’s wrong. A session where you struggle is a session that worked.
Do I have to pay for it?
No. Agora is free for students in the courses using it now, and anything your course assigns is free for everyone in the room.
What happens to my voice?
Agora doesn’t keep a recording. Your voice streams to the AI models that run the live conversation, and what Agora stores afterward is the text transcript. The companies that provide those models aren’t allowed to train on your sessions, and Agora doesn’t sell your information or use it for advertising. The details.
Could I get through a session without studying?
You can try. It’s a live spoken conversation and Glo’s next question comes out of what you just said, so a borrowed answer survives the first question and falls apart on the second. Faking it is genuinely more effort than knowing it.
My course doesn’t use Agora. Can I sign up anyway?
Not yet. Agora works because it’s built from your professor’s own materials and the ideas your exam will actually come from, so a course has to be set up first. If you want it in a class you’re taking, the fastest route is telling the person who teaches it.
“An AI did the assignment. Another AI graded it and gave feedback on it. Every box got checked, every deadline was met, the gradebook filled itself in. No human learned anything.”
Agora is built by someone who has watched AI hollow out homework from both sides of the classroom, as a student and as a teaching assistant.
Read why I’m building Agora
Find out this week, not on the exam
If your course is on Agora, this week’s review is already waiting for you.
Hear from every student, every week
You have never been able to hear a hundred students a week. Not because you didn’t want to: there are a hundred of them, one of you, and no hour in the week where that arithmetic works out.
The week as it runs now
- You ask if everyone follows. The room nods. Nodding and knowing look identical from the front.
- The signal you get is the three students at office hours, and homework that may have been written by a machine.
- The real picture arrives with the midterm, by which point the class is three chapters downstream of the problem.
The week with Agora
- Every student explains the week out loud, including the ones who have never once spoken in your class.
- You shape what the review asks before it goes out, and what you tell Glo becomes a rule it follows, not a suggestion.
- By the next morning you know which ideas landed, which are shaky across the room, and what is worth ten minutes before you move on.
You never open an individual student’s session, and that is the deal that makes them honest inside one. What reaches you is the class: where the week landed, who took part, and the few students worth a word this week, with a reason attached.