One model pass, streamed live. R0 lights before any streamed model text is read; later labels light as the corresponding sections of the result stream back.
Stage: idle
Your verdict.
Run details
--/100Reading
Scored on evidence, not potentialEvidence-led score
Why
The verdict appears when the scan is done.
AI scores have a few points of natural variance. The verdict above is more stable than the exact number; the Missing evidence list below is the most actionable signal.
Score breakdown
Next step
The next step appears last.
Decision
Missing evidence
Verified open jobs
Filtered from Job Radar by inferred city and role category. These are verified open roles, not a fit score or guarantee.
Risky claims
What an interviewer might push back on. Read it as preparation, not a list of mistakes.
This is why we do not just trust the model.
Example rewrite
BeforeWaiting for a weak bullet.
AfterWaiting for the safer rewrite.
Three questions, ten seconds
Nothing here is stored with your CV or JD. It only tells us who this is for.