How can a candidate take AI-Assisted Coding assessment?

What this assessment is

This is a coding assessment where you solve problems in the usual code editor, with an AI assistant available beside you while you work. The assistant is part of the test — using it is allowed, expected, and assessed. It is not a trick.

Two things are measured, separately: your coding result, scored the way coding tests normally are, and how well you worked with the AI assistant.

Your screen

The test screen has three areas: the problem statement on the left, the code editor in the centre with run and submit controls, and the AI Assistant panel on the right. You can collapse the panel to focus on your code and expand it whenever you need help. The timer, section navigation, and question navigation work exactly as they do in any iMocha coding test.

What the assistant can help with

As the panel itself tells you when you open it, the assistant can help you understand the problem and debug your approach, but it won't solve the problem or write the solution for you. Specifically, it can:

Explain compiler errors, runtime errors, and unexpected results.
Clarify programming concepts, algorithms, and syntax.
Discuss solution approaches and their trade-offs.
Suggest what to investigate or try next when you're stuck.

It won't write the solution, produce code that passes the problem, tell you whether your answer is right or wrong, discuss how the test is scored, or answer off-topic questions. This is by design — it is built to help you think, not to think for you. If it declines something, that is normal behaviour, not a bug.

For debugging, it explains what an error means and where to look, rather than rewriting your code.

Sharing your work with the assistant

The assistant already has the problem statement — ask your question directly. But it cannot see your code, your run results, or your screen. It only knows what you share in the chat.

To give it something concrete to work from, type @ in the message box and attach:

My current code — your code as it stands right now.
Test cases — the test cases along with their actual output.

For example, attaching your test cases and asking why they are failing gets you far more targeted guidance than asking in the abstract.

Anything you attach goes only to the assistant as part of the chat. It does not change the problem or count as a submission.

How you are assessed

Your coding is scored normally. Separately, the assessment looks at how you worked with the assistant — in plain terms:

Explain your problem clearly. Share the specific code, the exact error, and what you expected versus what happened.
Think for yourself. Form your own ideas, try things first, and question what the assistant suggests rather than just asking for the answer.
Build on the help you get. Act on hints and refine your approach across the conversation instead of starting over each message.

Some things worth knowing:

Using the assistant does not lower your score. Using it well helps.
Your AI result does not depend on whether your code passed. The two are scored independently.
Quality beats quantity. There is no penalty for asking several questions. What matters is how you ask and what you do with the answers, not how many messages you send.
Not using the assistant means there is nothing to assess on the AI side. Your coding is still scored, but engaging with the assistant is part of what this assessment measures.
You can chat in your own language. You will not be marked down for the language you use.
No prior AI experience needed. You type in plain language, like a normal chat. There is nothing to learn upfront.

There is no cap on the number of messages. Each single message is limited to 15,000 characters, shown by the counter in the message box, so an unusually large paste may need to be split.

If something goes wrong

The assistant isn't responding or the panel won't open. Check your internet connection and try again after some time. Your submitted answers are saved. If it still doesn't work, contact support with your test name and the time it happened.
You don't see an AI assistant at all. The assistant only appears on tests where it has been enabled. If it is not there, take the test as a standard coding assessment.
You didn't finish all questions. Submitted questions are scored: code and, where you used the assistant, AI fluency. Unreached questions are simply not scored.