Student guide
A challenge is a conversation where your questions, evidence, revisions, and explanations are the work. This guide covers each screen in the order you use it.
How to use this guide
This guide is divided into numbered workflows. Each workflow is a task made of numbered steps, and workflows appear in the order you first need them.
Workflow 1
Your teacher gives you a class code (it looks like XXXX-XXXX) or a join link. Open the portal and choose Create account, then create your sign-in with your email address and a password in the same tab. Back in the portal, Finish setting up your account asks for a display name, your date of birth (you must be at least 13), and the Class code from your teacher. A join link fills the code in for you. The code is required in the beta: it places you in your teacher's class, and your assignments appear on the dashboard. If your school set up your account for you, skip this and sign in below.
Open the address from your teacher or institution and choose Sign in on the portal card.
Sign in takes this tab to the sign-in form. Enter the email address your school or teacher used to create your account. When sign-in finishes, this tab returns to the portal.
Sign in with the temporary password from your administrator. The sign in form asks you to set a new password, and the temporary password then stops working. If you did not receive one, select Forgot password? and use the emailed one-time code. No one in the portal can see your password.
On later sign-ins, enter the password you chose in the sign in form. If you forgot it, choose Forgot password?; the portal cannot reset it.
This tab returns to the portal and loads your dashboard. Nothing needs to be clicked in between. Once the header appears, a spinner beside AI Challenge and the environment badge indicates activity while API calls, page loading, navigation, and other async actions finish. The dashboard reserves space beside its cards for the scrollbar.
Read the privacy policy and terms of use, tick both boxes, and choose Agree. You are asked again when either document changes.
Workflow 2
The dashboard lists your assignments under "Challenges". Each assignment keeps its progress on its row.
Each row shows a type, status, progress, and action. The status can be Not started, In progress, or Completed; dates can read Opens, Due, or Closed. A canceled assignment reads Cancelled, and its stopped session reads "Cancelled by your teacher"; Review on it opens the conversation so far. A run that ran out of time offers Review, which opens its report. A run you paused reads Paused. Some challenges have an age rating for mature topics. If a row carries a line such as Not available (age 16+) and its start button cannot be chosen, the next challenge in it is rated above your age, or no date of birth is recorded on your account, and there is nothing you can start there. Ask your teacher if you think your date of birth is missing or wrong.
Start challenge begins new work, Resume reopens a current run, and Review opens its report. The portal blocks work outside the Open At and Close At window.
Search by assignment title. Include completed and Include expired show hidden work.
The Your progress card on the dashboard shows only your own progress; nobody is ranked against you. It reads your level with its skill name, for example "Level 2: Questioner", your XP and how much more you need for the next level, mastery per challenge type, your badges, and a suggested next challenge type and difficulty. A graded solo run that finishes (Completed, partly completed, or out of time) counts; a team run and an abandoned run do not. XP comes from the credit on each required step: by default 10 XP for a full-credit step, more for a harder challenge (1.5 times at Intermediate, 2 times at Advanced by default; an unrated challenge counts as Introductory, and the rating is read when the run is counted) and for a harder support mode (1.25 times in Standard, 1.5 times in Challenge by default). Levels go from 1 (Curious) to 8 (Sage) and never go down. Mastery for a type is your average step credit over your latest runs of that type (the last 5 by default). Badges are earned once per challenge type: First mastery for your first Mastered run of a type, Challenge mode finisher for completing every step in Challenge mode, No hints needed for completing every step without a hint, and Five runs for five graded runs of a type, whenever you do them. The suggestion picks the type where your mastery is lowest, one difficulty up after a Mastered run and one down after a Beginning run. Under Evidence by skill and standard the card lists, under each framework's name and followed by its copyright or licence notice when it has one, each thinking skill your graded runs measured and each standard your challenges are aligned to, for example "questioning: 72%, Proficient (3 observations). Asking questions that clarify terms, probe assumptions...": the skill or standard's code, your average over your latest observations of it (the last 5 by default), its level when the framework has levels (never on an assessment framework, where it would read as a predicted result), how many observations there are, and the start of its description. A skill is scored from the evaluation dimensions that measure it; a standard from your step credit on a challenge aligned to it. It is evidence toward the skill or standard from your own work, not a test result.
The list shows ten assignments per page. Prev, Next, and the page label appear when needed.
If you open a teacher's join link when your account already exists, the dashboard shows "You have a class code waiting from your teacher's link" with a Join a class button. Choose it to join that class too; see Joining another class.
Workflow 3
Replies from the AI can take up to two minutes during beta. After about 20 seconds the page adds the text "The AI is still thinking. Replies can take up to two minutes during beta." under the response box; your message is kept if a reply times out. If the AI service's content filter blocks a message, the run pauses and the message is passed to a person to review; the page reads "Your last message was blocked by the AI service's content filter and has been passed to a person to review. Choose Resume to continue. If something is worrying you, talk to someone you trust." Your message comes back in the box, and you can choose Resume and carry on. If the filter blocks a challenge's opening, the challenge does not start and the dashboard says so: try another challenge or tell your teacher.
Start challenge creates and opens the session. If you already have that challenge open, you go back to it instead of starting it again. If the AI facilitator is busy or still preparing your challenge, the dashboard shows "Your challenge is being prepared - trying again in N s..." and tries again by itself; your challenge is kept and no attempt is used. If the AI is still busy after three tries, the dashboard reads "The AI is busy right now. Please try again in a moment."; choose it again shortly. If it is still being prepared after about 2 minutes, the dashboard reads "Your challenge is still being prepared. Open it again in a moment." and stops trying; it does not open the challenge by itself, so choose it again shortly.
The panel at the top of the conversation explains the challenge type and, under it, The scenario the challenge is about (Hide collapses it); the agent opens by stating the scenario, your goal and the rules of the challenge type in plain words, then asks its first question. The status shows In progress, and each turn names its speaker. The How this challenge works panel names the type and links to its full requirements and scoring reference. A challenge with a time limit shows the minutes allowed and the minutes used under its title; the minutes used update each time you send a turn rather than ticking, and when the limit is reached the challenge ends and is scored on the work done.
Under the challenge title the run shows its support mode. Your teacher picks it for each assignment, and it is Guided by default. In Guided mode a Your steps panel above the conversation lists the steps of the problem and marks each Done or Not yet as you go. In Standard mode the steps stay hidden until your report, and you can still ask for hints. In Challenge mode there are no steps shown and no hints. Unless your teacher has turned it off, a Standard or Challenge run offers Switch to Guided beside the mode; confirm with Yes, switch to Guided or choose Cancel. You cannot switch back to a harder mode for that run.
Enter your response and choose Send response. The Options page controls Enter-key behavior.
Your turn appears immediately as Sending... and the response box locks while the agent replies. The conversation keeps its newest turn in view, then the confirmed reply appears beneath yours. If the platform is busy, a note under the conversation reads "Busy right now - sending again in N s..." and your message is sent again by itself. If it is still busy after three tries, the note reads "The AI is busy right now. Please try again in a moment." and your message goes back in the box unanswered; send it again shortly. If an administrator has paused AI sessions during a provider outage, that is a different message: it says so, nothing is sent or tried again, and your message goes back in the response box; try again in a couple of minutes. AI work can also pause for the rest of the day when the platform reaches its daily AI spend limit; the message then says so, and it clears at midnight UTC. Nothing you wrote is lost. Starting a challenge shows the same message. If a reply is slow, the page reloads your session; your message is not sent twice. If the reloaded session already has your message, it stays in the conversation; if not, it goes back in the response box. "Still answering your last message..." means an earlier send, from this or another tab, is still being answered. To send a message that went back in the box, send it unchanged: the platform recognises it and never answers it twice, while an edited message counts as a new one. If the agent cannot reply at all, the session pauses with "This session is paused after a problem on our side"; your message is kept, and after you choose Resume you send a new message to carry on.
You get two attempts at each challenge. Going back to a challenge you already have open does not use another attempt. If you leave an assigned challenge unfinished twice, the assignment moves you on to the next challenge. You can start up to five new challenges a day (the day resets at midnight UTC). Picking up a run you already started does not count toward that. When you reach a limit, the portal tells you so: "You have used all attempts for this challenge." or "Daily challenge limit reached. Try again tomorrow."
Your teacher also sets a turn limit on each assignment: the most responses you can send in one run. The run screen does not count your turns down. When you send the last one allowed, the agent gives a closing reply, the challenge finishes, and it is scored on the work so far, the same as a challenge that finishes on its own.
Types appear on library cards, assignment cards, in the run screen header, and in the How this challenge works panel at the top of every session. Each type runs the conversation by its own rules and is graded differently; see the challenge types reference for what each one asks of you.
Explain why evidence matters, name uncertainty, and revise your view when the conversation gives you a reason.
Workflow 4
Your session saves as you work. Choose Pause beside Send response to stop on purpose: the response box is replaced by a PAUSED panel with a Resume button. You can also choose Dashboard or close the tab without saving.
Choose I'm stuck beside Pause. A note reads "Hint on its way - send your next message and the reply will include it."; the agent's next reply carries the hint, and each hint you ask for counts toward your hint total. The button refuses with a message when no more hints are available on the challenge, when a hint is already waiting for your next message, when a timed challenge is nearly over, once the assignment has closed, or on a team challenge where your team answers through proposals. In Challenge mode the button is not shown.
The row reads In progress, or Paused if you paused it, shows your turn count, and offers Resume.
Resume restores the conversation and response box where you left off.
Workflow 5
The agent ends the reasoning session. The response box becomes a result panel.
Your report appears a few seconds after the session ends: until then the panel reads "Your report is being prepared...". If it takes longer than about 2 minutes, the panel says so and the report appears on your dashboard when it is ready; you can leave the page. A full report opened before the score is ready says the same and fills in by itself once it is. In the rare case that the report cannot be prepared, the panel says so; your conversation is still saved, so let your teacher know. The panel then shows a score and summary, with a See full report button. Use it to open the full report: your score, a written summary, dimension scores with a rationale for each, what to practice next when the evaluation named something, and the transcript.
A challenge is made of steps, the parts of the problem you work through. Each step earns credit on its own, so finishing part of a problem still counts: full credit for a step you complete without a hint, a little less for each hint level it took, and some credit for a step you made a real start on. The report heads the score with your mastery band, for example "Developing - you completed 3 of 6 steps; 2 with a hint.", and lists every step as Completed, Completed with hint level N, Started or Not yet. The bands, from first steps to full command, are Beginning, Developing, Proficient and Mastered. The report also shows the support mode the run ended in; your teacher sets the mode for each assignment.
For a graded solo run that finished, the report says how much XP the run earned, for example "This run earned 38 XP.", and lists any badge it earned. Progress is added a moment after the grading, so a report opened before then says your XP is still being added and keeps checking for about 10 minutes, showing the XP as soon as it is added; after that it reads "Your XP for this run will be added shortly." A run whose progress was missed, or one graded before progress existed, is added by a background check that runs every 30 minutes. A run that does not count (a team run, an abandoned run, or one that could not be evaluated) shows no XP line. See Track your progress.
A finished challenge is marked Completed and offers Review, which reopens the full report. Choose Include completed if the row is hidden. A challenge that ends Abandoned after you sent at least one response still gets a report, graded on the work done before it ended, and offers Review once it is ready. One left before you answered at all has no report, so its row offers no button.
Open the activity from your course. Depending on your institution, it opens within your course or in a new portal window. If the activity is no longer available, read the launch notice and choose Continue to AI Challenge to open your dashboard. If a launch inside the course page says "Open the activity in a new window or tab", open that activity in a new tab from your course to complete sign-in. Reloading or opening another portal tab keeps your course session while it remains valid. Reloading a course launch returns you to your dashboard; open the activity from your course again to return directly to that assignment. Brief connection failures retry automatically. If another course activity takes longer to sign you in, the older course tab retries for about 15 seconds, then asks you to reopen the activity if sign-in has not finished. If another tab in the same browser signs in or starts a course activity, your open portal tab reloads and switches to the newer sign-in. If another tab signs out, your open portal tab reloads and shows sign-in. If the portal says your course session has ended, open the activity again from your course. A later direct visit with an expired course session returns to normal sign-in without that message. Course launches are disabled by default during beta. A launch session cannot renew beyond eight hours; reopen the activity from your course after it ends. Your school can send overall, dimension or completion grades back to its LMS. If you use a partner platform with its own interface, follow that platform's instructions; it uses the partner API rather than these portal screens.
Workflow 6
On a wide screen the panel stays open on the left; on a narrow screen choose the menu button. The panel contains Dashboard, Notifications, Feedback, Join a class, Options, About, and Sign out. The bell in the header also opens Notifications and shows how many are unread.
If a teacher gives you a code for a second class, choose Join a class, enter the code, and choose Join. The page confirms the class you joined, and its work appears on your dashboard.
Choose light, dark, or system theme. The choice is remembered.
Choose whether Enter sends or starts a new line.
Your school manages your account and your work, so there is no delete button for students. To have your account and personal data erased, ask your teacher or your school's administrator; they can erase it from the admin portal. Erasing is permanent. See the Privacy Policy for what is erased.
Choose Sign out once. This tab returns to the sign-in card with "You're signed out." The portal also signs you out after a period of inactivity (60 minutes by default, set by the platform or your institution). A warning appears two minutes beforehand with a countdown and Stay signed in. Typing counts as activity, and activity is shared across tabs. Opening another tab keeps the existing inactivity deadline. Signing out of the student portal signs out its other tabs. A new sign-in ends the previous student sign-in session. If the inactivity setting cannot be loaded, you can stay signed in for up to 12 hours until the next successful refresh. Unsent responses are lost on sign-out and are never saved in browser storage. Use it on a shared machine.
Challenges can carry labels, short topic words. They appear on the cards of the challenge library, which is shown to independent learners only (see the independent learner guide for the filters). If you joined a class, your dashboard lists your assignments, and they do not show labels.
Internal agent instructions stay hidden. The details of each step stay hidden during an active challenge unless it runs in Guided mode, which lists them as you go; every run lists the steps on your report once the challenge is graded.
Scheduled announcements appear above the portal header, with Info, Warning or Critical severity and Critical announcements first. Your teacher can create them for students in rosters they own; institution administrators can reach their own institutions and narrow by role; platform administrators can reach everyone or selected roles, institutions and rosters. You see only announcements whose audience you match, including every selected target kind. Students cannot create banners.
The author schedules start and stop dates in their browser's local time. Banners refresh when you sign in or load the app and every five minutes while it is open. Close a banner to dismiss it while this app remains loaded; a fresh page load can show it again if its schedule is still active.
Loaded announcements appear and disappear at their scheduled times even when a refresh fails. When several announcements are visible, scroll within the banner area to read them; the header remains below the bounded area.
Assignment notifications are queued with the assignment and may take a few seconds to be created. The bell checks for unread notifications about once a minute. Each browser starts at a random moment and varies the interval by up to 20%, so a whole class does not check at once; the unread count can therefore take up to a minute to appear after the portal loads. It does not check while the tab is hidden, and checks again when you return to the tab if a check is due. Brief connection failures keep the last known count, and while checks fail the portal waits longer between them (up to 10 minutes) and respects a wait the platform asks for. After three checks fail to reach the platform without a successful check between them, one message says notifications are temporarily unavailable; the portal keeps trying automatically. Opening the Notifications page marks everything it shows as read in a single request. By default, a notification is removed 90 days after you read it, or 180 days after it was sent if you never read it.
Request allowances apply across the whole service. If you reach a limit, wait for the time shown before trying again. If AI requests are temporarily unavailable, wait and try again when the service recovers.
