Once your AI-scored interview ends, it turns into a document a recruiter will actually read. Knowing roughly what's in that document changes how you'd think about the round itself, since it's not just a pass or fail number.
The scorecard itself
Most platforms build the report around the specific competencies the role is being scored against, giving each one its own score rather than a single overall number. Each of those per-skill scores is typically backed by a citation back to the transcript, the specific part of your answer the system used to arrive at that score. Some platforms attach a confidence indicator to each assessment too, essentially a signal of how reliable the system considers its own read of that particular answer, which matters because it tells a recruiter which parts of the report are worth double-checking themselves.
Summary, flags, and proof
On top of the raw scores, most reports include a written summary highlighting your qualifications, relevant skills, and any areas that came up short, so a recruiter can scan the summary instead of replaying the full recording. Red and green flags mark specific moments, green where an answer clearly supported a competency, red where there was a gap, a contradiction, or a response that missed what was actually asked. What makes these reports checkable rather than opaque is that every score generally links back to the exact timestamped clip or transcript line it came from, so a recruiter looking at a borderline case can verify the score themselves instead of taking the number on faith.
What this means for you
Because the report is built around specific, citable moments, a single strong, well-structured answer to a key question can carry real weight, it becomes the clip the system points to as evidence. The flip side is also true: a rambling or off-topic answer to an important question becomes an easy citation for a red flag. The report is an input that helps a recruiter decide who to look at first and what to focus on if they follow up, not a locked verdict, a human is still the one reading it and making the final call.