Skip to main content
A Round is one step inside a Role. If a Role is the whole journey, a Round is a single door the candidate walks through. You can have one Round (a quick screen) or many Rounds in a row (screen → technical → final). Each Round runs its own interview.

​
How Rounds connect

Each Round has its own Interviewer (the AI persona for that step) and its own pass mark. You create the Round and its Interviewer together, in one step — there’s no separate “create an interviewer” beforehand.

​
Every field, in plain words

​
uuid
required
Which template this stage belongs to.
​
string
required
A short label for the step, like "technical".
​
string
The name the candidate sees, like "Technical Screen".
​
enum
The format: ai_interview, roleplay_simulation, practice_session, manual_review, async_assessment, or final_review.
​
object
required
The Interviewer that runs this step, defined inline — created together with the Round in one atomic call. Contains the interviewer’s name, persona, evaluationDimensions (weights sum to 100), additionalInstructions (the single AI-instruction field), and optionally context, interaction, voiceGender, aiAvatarId. New interviewers default to a 0–10 score scale. See Interviewers for every field.(To reuse an interviewer another Round already has, pass interviewer_id (uuid) instead of this object.)
​
integer
The position in line. 0 is first, 1 is next, and so on.
​
integer
How long the interview lasts (default 30, max 60).Guidance: if coding_enabled is true, set this to around 50 minutes — a coding round needs time to read the problem, write and run code, and discuss the approach. For any other interview round, use a minimum of 40 minutes; below that there isn’t enough time to cover the rubric properly. Phone-screening rounds are the exception and stay short.
​
number (0–100)
The score needed to pass this step.
​
boolean
Can the candidate try again if they fail? Default false.
​
enum
What happens when the step finishes:
  • auto_advance_if_passed — move on automatically when the candidate passes
  • require_reviewer_approval — wait for a human to say yes
  • auto_reject_if_failed — reject automatically on a fail
  • auto_assign_next_stage — assign the next stage automatically
  • none — do nothing automatic

​
Build a stage, step by step

Stages live inside a template, so you add them from a template — in the dashboard or over the API.
1

Open the Role

In the left sidebar, click Roles and open the one you want to add a Round to.
2

Add a Round

In the Role’s Rounds section, click Add round.
3

Name it and pick the format

Set the Title and type (AI interview, roleplay, practice…).
4

Set the order and pass mark

Set the order and the pass threshold, then choose what happens on completion (auto-advance or wait for approval).
5

Define the Interviewer (step 2)

The form’s second step is the Interviewer that runs this Round — its name, persona, context, behavior, rubric, and AI instructions. It’s created and linked when you save.
6

Save

Save the stage. Repeat to build a multi-round pipeline.

​
Examples for each use case

These examples show the round shape and pipeline only. Every round is created with its own interviewer object (see the create example above) — it’s omitted here for brevity, but it’s required on each POST /api/v1/rounds.
Several steps, a rising bar, and a human check before the last round.

​
Next

Participants

Add the people who will go through your Rounds.

Create a Round (API)

The full endpoint reference.
Last modified on July 26, 2026