Meeting Summary

After a meeting, Copilot can review it: which questions were asked, how you answered, what you missed and what to work on. The summary opens in the meeting preview on the "Summary" tab.

Getting a summary

Click "Generate summary" in the meeting preview or the icon next to the meeting in the list. Generation runs on the server, so you can close the page and come back later.

• The summary is written in the interface language. If you change the language later, the existing summary isn't translated; generate it again.
• The summary doesn't update itself. If you continued the meeting, click "Regenerate". The new summary fully replaces the old one.
• The model selected in the chat doesn't affect the summary.
• For recordings uploaded in the web dashboard, the summary is created automatically.

What's inside

The summary opens with a short overview and the interview type: technical, System Design, behavioral, HR or mixed.

"Overall Performance" shows the success rate and average scores for correctness, completeness and clarity. The success rate is the share of answered questions with an average score of 6 out of 10 or higher, so questions you didn't answer don't lower it.

Next come strengths, areas to improve and recommendations. If the meeting has a vacancy with requirements, "Vacancy fit" appears next to them: which requirements you demonstrated in your answers, where the gaps are and what the meeting didn't touch. They're followed by the "Competency map" with the topics you were tested on. Each topic is rated "Strong", "Medium" or "Weak" based on your answers, not on the topic merely coming up.

In the questions section, each question has three scores, your answer, feedback, what was missed and a reference answer. The "Voice", "Text", "Screenshots" filter only changes what's displayed.

In behavioral and HR interviews the scores mean something else: correctness means relevance and plausibility of the answer, completeness covers the STAR structure with a measurable result, and clarity rates delivery.

What the summary is built from

The summary covers all subchats of the meeting, including what happened before a context reset: the interviewer's transcript, your transcript and messages, screenshots and the assistant's answers. Only what you said yourself is scored: the model sees the assistant's answers but doesn't credit them to you. The context, notes and video recording aren't used. The vacancy doesn't affect the scores, so meetings stay comparable with each other: only "Vacancy fit" is checked against it. The block is built only if the transcript contains your answers; older summaries get it after "Regenerate". Greetings, connection checks and repeated questions don't make it into the list.

Deleted messages are left out of the next generation; the existing summary doesn't change.

Export and history

• The "Export" menu on the "Summary" tab saves the summary as PDF or Markdown: as a full report or only the questions with their review. Screenshots aren't included in the file, only their count.
• The "History" tab shows the meeting's entire conversation in order and lets you search it. It can also be exported to PDF or Markdown. The "my messages" option includes both sides' transcript and your messages, without the assistant's answers.
• The "Recordings" tab shows interview recordings of this meeting made on this computer.

Search across all meetings opens with Ctrl/Cmd + K. It finds meetings by title and messages by text, but doesn't look inside summaries.

Add to bank

You can publish the questions from a summary to the public bank of interview "leaks" and get free subscription hours: 6 for HR, 12 for technical and final rounds, 24 for System Design. The hours are credited even if you don't have a subscription right now.

Only the question texts are published, with names, contacts and links removed automatically. Answers, scores, the transcript and the author aren't published. A publication can't be withdrawn.

Publishing fails if:

• this meeting has already been published or was created less than 30 minutes ago;
• in the last 24 hours there were already three publications or 24 hours were credited;
• fewer than three meaningful questions remain after cleanup;
• you've already published most of the questions before.

Tracker and mock interviews

If the meeting is attached to a card in the application tracker, the card shows the average success rate of its meetings, weak topics and recommendations. After you regenerate the summary, the numbers update right away. The summary itself shows a chip with the company and stage: from it you can open the card or move it to another stage.

"Start mock" opens a mock interview built from the summary: with your role, grade and a focus on weak topics. You can choose new questions based on gaps, a repeat of low-scoring questions, or a mix.