Get a session
GET
/v1/sessions/{id}Scope results:readstatus is created, in_progress or finished. durationMs is set once it ended. endReason says why a finished session ended: completed (every block ran), candidate_ended (the candidate ended it early) or abandoned (they left and it timed out); endedEarly is true for the last two, so the results cover only the blocks they reached.
Response
Get the results
GET
/v1/sessions/{id}/resultsScope results:readWait for the results.ready webhook, or poll until analysisStatus is done or error: until then, blocks appear one by one, are pending or running, and the scores can be null. Reviewers' corrections in the dashboard are already applied.
Response
Results fields
| Field | Type | Description |
|---|---|---|
overallScore | integer or null | 0 to 100: how much of the interview's expectations the answers covered |
mustHavesMet | boolean | Whether every must-have expectation was covered |
analysisStatus | string | The analysis as a whole: pending, running, done (final, results.ready was sent) or error (failed; your team can retry it in the dashboard). Poll on this, not on the blocks |
endedEarly | boolean | The candidate ended the interview before every block ran, or left it |
decision | string | Your team's decision in the dashboard: undecided, advance, hold or reject |
summary | string or null | The AI's summary for the reviewer |
followUps | array | 5 to 8 suggested questions for a live follow-up interview, each with rationale, basedOn and evidenceRefs; empty while the analysis runs |
blocks[] | array | Per block: status (pending, running, done, error), score, summary and coverage |
blocks[].coverage[] | array | Per expectation: covered, partial, covered_after_prompt, not_covered or incorrect |
consent | object or null | What the candidate agreed to before starting: when, web or embed, and each text's version and sha256. identityVerifiedBy is company when the invitation had skipSignIn (your site verified the candidate), else null |
label | string | The reminder to show with the scores: for review only, not a hiring decision |
resultsUrl | string | The results page in the dashboard (sign-in needed) |
aiGenerated | object | Which fields our AI wrote (fields) and the IPTC source type (digitalSourceType): the machine-readable mark of the EU AI Act, Art. 50(2). Scores are computed from the statuses, not by the AI |
Note
The results never include integrity signals (tab switches, pastes), the recording or the transcript. The transcript is only in the PDF report with
?transcript=1; integrity signals and the recording stay in the dashboard.