A project's two conversations
The Creation Console and the viewer's chat, kept apart, and what each knows. The overview is Projects.
Two conversations, kept apart
Both of a project's conversations are about that project and nothing else. Ask one something unrelated — the day's news, say — and it tells you so in a line and points you to the Home chat, which is for anything at all.
A project with viewers has two kinds of conversation, and they never mix:
- The Creation Console (the middle of the page) is for the project's creators — the owner and contributors. This is where you direct the work, and the only place progress notes appear. Kuyun's own progress — findings, reviews, entries filed — is gathered into one line between your messages, such as "Kuyun worked · entries filed ×30"; tap it to see every step. The notes it writes while it works on something you asked for are folded the same way, into "Kuyun worked · N steps" above its answer. What needs you, like a question or a turn that failed, gets a line of its own. If you use Kuyun on two devices, a conversation you left open on one catches up with the other when you come back to it.
- The viewer's chat is what the people you share with get. Their page has the same shape as yours: the conversation in the middle and, on the right, what is in the project's Output folder, without the folder around it — the README, the deliverables, any map, the tables in Output and the files you shared with them — and nothing of the making-of, which their chat is not told either. Yours is the work's own page — ▶ View at the top of the project in Create (Shelf, "A work's own page") — as your own rehearsal; it names the files as your viewers see them, so "the README" there is the one in your Output folder. For a viewer it is simply the conversation, with no name of its own. On a course it is a study partner that teaches. On a coding practice it is a coding partner, and above the conversation is the list of problems: by topic, each with its level and a mark once you have solved it or tried it. Tap one to open it in the editor (Studying with a project, "Writing code"), narrow the list by topic, level or a word, or ask the partner to pick one for you; hide folds the list away. What you solved is yours alone to see. On a role-play practice it is a practice partner that plays the other person in each scenario — the barista, the interviewer — sets each stage of the scene in a line, gives a hint when you are stuck, and after the last stage steps out of the role to say which goals you met, what you said well and what to say instead. You can type or use the microphone; with the microphone it judges the words it heard, not your accent. Above the conversation a scene card shows where you are: the picture of the stage you have reached — always whole, made smaller to fit your screen rather than cut — and a line saying what is happening, with the situation, your goals and the useful phrases folded underneath — tap them open when you need them. It moves on as the partner moves to the next stage, and ▴ hides the picture to give the conversation more room. ▶ Start the first scenario (and ▶ Continue, ▶ Next, 🔁 Play it again) turns on conversation mode — for that visit, even if you had turned it off before — (🗣, beside the microphone, turns it on and off — Talking instead of typing): each reply is read aloud as it arrives, the microphone opens by itself when it ends, and pausing — or tapping the bar — when you have finished sends what you said, as a voice message. So the partner speaks first and the microphone is waiting when it finishes. While the mode is on, the typing box and the microphone button give way to one bar that shows whose turn it is (Talking instead of typing); to type a line, turn 🗣 off. The first time you type a line of the scene the partner asks you to say the next one aloud, except in a round that asks you to write code: what you paste is your answer, read as an interviewer would read it, so turn 🗣 off to paste it. When it could not make out a reply — noise, a fragment, words that fit no stage — it says so, repeats the other person's line, and after a second miss gives you a phrase to say back; it never moves on until it hears a reply that belongs to the stage. Say nothing for ten seconds and the microphone closes, and the bar says Tap to talk; 🗣 turns the mode off, and leaving the page closes the microphone. Nothing is recorded while the partner is speaking, and a pause in which you said nothing sends nothing. The partner says each line once, in the language you are practising — never with its reading or a translation after it. Under each of its lines are ▶, to hear it again — the very recording it was said in, or, for a line that was not read aloud, one made the first time you tap it and then kept, so every tap after plays the same thing — and 译, which shows the line in your language in a small window over the page; tap anywhere outside the window and it goes away. In a role-play they are always there, even while the translate and listen buttons are switched off everywhere else. Your own turns have neither: they are your words, so the one button under them is the ▶ that plays back your recording. At the end of a scenario the partner's feedback comes as a card: each of the scenario's goals with ✓ if you met it and ✗ if not yet, what you said well, what to say instead next time, and how many of the scenarios you have practised so far. The count is your own; nobody else's practice is in it. A scenario's file, with the model answers, opens once you have practised it, as a course's lesson does; the card shows nothing of those answers. The assistant will teach any chapter you ask for, in any order, and every chapter is also open in the files from the start. In a course or a role-play practice, tapping a lesson or a scenario in the files, or its practice, starts it with your partner here in the conversation ("Let's start Lesson 1 — …") rather than showing you the file, which would only have shown its answers out of turn. The reader (📖) still has them. A course or a role-play practice can keep its files shut instead (Lock what a viewer has not finished yet, in Publish). A course is open unless the person who made it ticks that; a role-play practice is locked unless they untick it, and they can change it there at any time. Then a lesson's file, with its practice and their answers, shows 🔒 and opens only once you have finished that lesson with your study partner — it records it finished, never just because you ask to see it — and a practice set for the whole course, such as a final test, opens once every lesson is finished; a scenario opens, with its goals and what a good answer looks like, once you have practised it. Neither you nor the study partner can open a locked one early. What you have finished stays open if the maker turns the lock on later, and everything opens if they turn it off. An archived course has nothing locked. What Kuyun keeps track of is where you were last (that chapter is marked "you are here" in the files) and which chapters you have finished (marked ✓). "You are here" appears only once you have opened a chapter — with the assistant, or yourself in the reader (Studying with a project): until then you are at the start. Nothing is graded — the assistant gives feedback on each answer as you go, records a chapter finished when you complete it, speaking practice included, and offers the next. Once a course is archived it stops doing either: the chapters still read and it still answers about them, but nothing is ticked off, there is no next one to go to, and "you are here" stops following you. Each viewer has their own conversation, and it is theirs alone: you cannot read it or join it, and neither can anyone else. Their first message is the study partner's greeting with a few ways to start under it, chosen for the kind of work — on a course ▶ Start lesson 1, What will I learn? or Quiz me to find my level; on a book Start reading, Start listening when it is narrated, or What's it about?; on an article Give me the summary. A chip that asks something sends those words as their own message, and the partner answers; reading and listening open on the page. Typing anything works just as well. The buttons keep coming after that: when a viewer comes back after half a day or more, the first time that day the partner greets them with where they were and buttons to continue at that lesson or chapter, or what fits the kind of work — today's briefing, what changed since they were last here; and right after they finish a lesson, a chapter or a scenario, buttons for the next one and what usually follows it — a quick quiz, the same scenario again, "what just happened?". When the people who made it have linked other works to it, those come too: after lesson 3, Play the scene opens the role-play that goes with it — or, in a programming course, Practise it opens the coding practice topic that goes with it — and A story the story written for it; after the last lesson, Next level; and after every topic of a coding practice, Next, the work its makers put after it. When its maker wrote a placement test, a new viewer's first button is Find my level: the partner asks a few questions, one at a time, tells them the level their answers show, and starts them at the lesson that fits, so everything that continues begins there. They can ask to find their level again at any time. The button under that answer starts them there. In a collection of works, the first buttons include Start, its first work at its first lesson, so someone arriving can begin at once; when its guide lists or suggests works, a button to open each one follows the answer; coming back offers to continue in the work they were last in, right at their lesson or chapter, and its placement can start them in any work it holds. A button that continues opens the reader or the player right at that place, or starts that lesson with the partner. They are drawn from where that viewer really is, so they change as they go. On an archived course the greeting says so, and only reading and listening are offered. A course has no schedule: nothing arrives on a clock, and a viewer studies when they like. When they finish their first lesson, the study partner offers them a reminder to study — every day, or once a week, at an hour they pick — and sets it only if they say yes. The greeting is in the language their own Kuyun is set to — English, Chinese or Spanish. A change to the project itself — the plan, the goal, a chapter's text, how the work is delivered — belongs in the Creation Console. How it is delivered is written in the Delivery notes, a file of its own beside Output that viewers never see: say how to teach, not what to teach — the chapters hold that.
Both conversations know where they are: the project's name and its ID, the link to its page, and who is speaking, in what role. Ask either one for the project's link and you get the real one. Anything about the page it has not been told, it says it does not know, rather than guessing.
▶ View is your own private run-through of the project, delivered to you exactly as a viewer would get it, at your own pace. Your rehearsal progress is yours alone — no viewer sees it or is moved by it. Once the course is archived there is no rehearsal progress either: the chapters still read, and nothing you do there is recorded.