Skip to main content
Rounds7 min read

Add an AI interview round

Walk the Create a round wizard. Name the round on Basic Details, pick the Interview type, set the round type and difficulty, fill the Questions step, and set the score cutoffs on Workflow.

Before you start

  • A saved job with a title and a job description.
  • Requirements set on that job, with a level on each row.
  • A view on which round type you need. The round type guide has the full table.
Steps

8 steps, in order.

1

Open Create a round and check the step count

Open the job and add a round. The modal is Create a round, and its left rail reads New round, Configure the interview round. How many steps you get depends on the job. With Enable workflow on, the rail is Basic Details, Round Configuration, Questions and Workflow, and the footer reads Step 4 of 4. With Enable workflow off, the rail is just Basic Details and Workflow, and the footer reads Step 2 of 2. Completed steps get a filled purple circle with a check.

  • Enable workflow is a toggle on the job, not on the round. Set it on Job details before you start adding rounds.
2

Name the round on Basic Details

Basic Details is the first step. A large round number sits on the left, beside Round title. Give it a name your team will recognise on the pipeline board. Coding Round. HR Screen. System Design. The name is what shows on the candidate table, so keep it short.

Dashboard›Clients›Stripe›Backend Engineer
Create a round
Basic Details
Set up basic round information
Round
1
Round title *
Enter round name
Interview type *
Resume Screening
AI analyzes skills, experience, and education to generate a readiness score - no interview required.
AI Interview
An AI interviewer conducts and evaluates the session, grading communication, confidence, and technical depth.
Human-Led Interview
A live meeting with the candidate - host it in Expert Hire, or on Zoom/Meet/Teams with a notetaker. Scored with a full report.
AI Prompt Assessment
Candidates demonstrate AI prompt-engineering skill by writing prompts to an AI tool for real-world tasks. Scored on 8 dimensions including prompt quality, context framing, and iteration strategy.
Coding Test
Candidates solve a sandboxed coding problem with real execution against hidden test cases. Scored on correctness, efficiency, code quality, and complexity.
BackContinue

Basic Details. Round title, then the five Interview type cards.

3

Pick the Interview type

Interview type sits on the same Basic Details step, as five selectable cards. Resume Screening reads resumes and returns a readiness score with no interview. AI Interview is the voice round with Ethan. Human-Led Interview is a live meeting, hosted here or on Zoom, Meet or Teams with a notetaker. AI Prompt Assessment scores prompt-engineering work on 8 dimensions. Coding Test is a sandboxed problem with real execution. Pick AI Interview for a voice round.

  • Interview type is the outer choice. Round type only matters once the Interview type is AI Interview.
4

Pick the round type on Round Configuration

Round Configuration is the second step. Round type is the flavour of the AI interview and there are eight: Coding, General Interview, Communication, HR Round, Problem Solving or Aptitude, System Design, Academic, and Task. The dropdown shows the length beside each name, for example General Interview 15 + 10 min, and an info chip above it reads 15 + 10 min · includes intro. Coding and Task open a code editor. System Design opens a whiteboard.

  • General Interview is the safe first round for a mixed role. It is 15 + 10 min and covers broad fit.
5

Set difficulty, non-negotiables and the desktop app

Difficulty Level runs from very easy to very hard and changes how hard the questions get. It does not change the length, which is fixed per round type. Two more controls sit on the same screen. Non-negotiables takes checklist questions one at a time, typed into Add a checklist question and added with the Add button. Require desktop app pushes candidates to take the round in the ExpertHire desktop app for stronger proctoring.

Dashboard›Clients›Stripe›Backend Engineer
Create a round
Round Configuration
Configure round settings and parameters
Step 2 of 4
Difficulty Level *
Medium
Round type *15 + 10 min warm-up
General Interview15 + 10 min
Non-negotiables
Add a checklist question
Add
No items yet - add one above.
Candidate warm-up (+10 min)
Adds a settling-in stretch before the graded questions. Leave it on unless the round is very short.
Require desktop app
Candidates are pushed to take this in the ExpertHire desktop app for stronger proctoring.
BackContinue

Round Configuration. Difficulty Level, Round type with its 15 + 10 min chip, Non-negotiables and Require desktop app.

6

Fill the Questions step

The Questions step has two blocks. Suggested questions is a checklist with a 0 / 10 counter and a Select all button. Custom questions is a textarea where you type your own, then press Add or Ctrl+Enter. A chip at the top counts what you have selected, and the empty state reads No questions selected. A Task round needs none of this. Its single challenge and its two or three follow-ups are generated for you.

Dashboard›Clients›Stripe›Backend Engineer
Create a round
Questions
Select round specific questions
Step 3 of 4
Select or add questions4 selected
Suggested questions4 / 10 Select all
What are the key differences between an ArrayList and a LinkedList in Java, and in what scenarios would you choose one over the other?
Walk me through how you would design a rate limiter for a payments API that has to stay correct across several servers.
How would you find and fix a memory leak in a long running Python service that only shows up in production?
Explain what a database index actually does, and describe a query you would not want to add one for.
When would you put a queue between two services instead of calling one from the other over HTTP?
How do you make a payment endpoint idempotent, and how would you test that it really is?
A read replica is lagging by forty seconds. How do you find the cause and what do you change first?
Describe a schema migration you shipped with zero downtime, and what you would do differently now.
How do you decide what to log, what to trace and what to alert on for a service you own?
Two services need the same reference data. When do you copy it and when do you call for it?
Custom questions0
Type a custom question...
Ctrl+Enter to add Add
BackContinue

The Questions step. Pick from suggestions, select all, or add your own.

  • Whatever you pick, follow-up questions still adapt to what the candidate says in the room.
7

Set the selection criteria on the Workflow step

Workflow is where you set score cutoffs, on a Score bands 0-100 slider. Anyone scoring in between goes to manual review. Auto-reject low scorers and Auto-advance high scorers are both off by default, so until you flip one every candidate stays in manual review. Each toggle carries an optional message editor. An empty Reject message means no rejection email is sent. An empty Accept message means the default shortlist template is used. Integrity is not part of this. A malpractice flag goes to a human reviewer. It never rejects anyone.

Create a round
Workflow
Selection criteria for this round
Step 4 of 4
Selection criteria
Set score cutoffs. Anyone scoring in between goes to manual review.
Score bands 0-10040 / 70
Reject below 40Accept above 70
Below 40 auto-reject, above 70 auto-advance. 40 to 70 goes to manual review.
Auto-reject low scorers
Reject candidates scoring below 40.
Reject message (optional)
Empty = no rejection email is sent. Fill in to notify the candidate.
Empty = candidate gets no rejection email.0/5000
Auto-advance high scorers
Advance candidates scoring above 70.
BackCreate Round

Score bands 0-100. Auto-reject and Auto-advance are off until you turn them on, and anyone in between goes to manual review.

  • Auto-reject low scorers is a score threshold you choose to turn on. It is never triggered by a proctoring or malpractice signal.
8

Create the round and invite candidates

Press Create Round. Add candidates to the job and send the invite. The candidate joins on their own schedule. An AI interview costs 5 credits, and proctoring adds nothing.

Dashboard
Good afternoon, Rachel
Your week: active clients, top candidates and upcoming meetings.
Clients62
View all
ClientActive jobs
Stripe6
Datadog3
Shopify2
Salesforce1
+58 more
Top candidates4
View all
CandidateScore
J
Jordan Blake
Senior Backend Engineer · Stripe
82
N
Nina Sandoval
Product Manager, Platform · Datadog
79
D
Dev Park
Senior Backend Engineer · Stripe
74
S
Simone Carver
Data Analyst · Shopify
68
Calendar
Scheduled interviews
TUE
7
AI Interview
Marcus Bell
AI Interview
Sam Cole
WED
8
AI Interview
Nina Sandoval
AI Interview
Sam Cole
THU
9
AI Interview
Grace Lin
AI Interview
Dev Park
FRI
10
AI Interview
Dana Whitfield
AI Interview
Grace Lin

The workspace after the round is live, with interviews scheduled against the job.

What you get

At the end of this guide.

A live round attached to the job, with a name that reads on the pipeline board.

A fixed length and the right tools in the room, set by the round type.

A question set, picked from the suggestions or typed as custom questions, with adaptive follow-ups on top.

Score cutoffs on the Workflow step, with both automations off until you turn one on.

Common mistakes

What goes wrong, and the fix.

Setting a round type without checking its length.

HR Round and Task are 10 + 10 min. Coding is 20 + 10. Match the type to how long you actually need.

Turning Auto-reject low scorers on for the first round of a new job.

Run the first batch with it off, look at the scores, then decide where the cutoff belongs.

Hand-picking questions for a Task round.

Task generates its own challenge and follow-ups. Leave the Questions step alone.

Expecting a malpractice flag to filter candidates out.

Flags go to a human reviewer. Read them on the report and decide yourself.

FAQ

Questions people ask on this one.

Still stuck? Talk to us and we will walk through your setup.

Read next

Related guides.

Rounds

Choose a round type

The five Interview type cards, the eight AI interview round types, their lengths, which ones open a code editor or a whiteboard, and when to reach for each.

8 min read
Scoring

Set requirements and levels

You pick the requirements for the role and set a level on each. The JD is scored against too. Questions adapt to the answer. The score comes back on the requirements you chose.

7 min read
Reports

Read a candidate report

The four tabs, the overall score, the skill bars, the recording and searchable transcript, the malpractice status, and how to share it.

6 min read
All guides

Try this in your own workspace.

The 30-day trial includes 75 Hire Credits, which covers a resume batch and a first round of interviews. Nothing charged for 30 days.

30-day free trial with 75 Hire Credits. Card required, nothing charged for 30 days.