AI Voice Interviews

Hivemind can run first-round phone screens for you. Drop a Voice Call stage into a pipeline and an AI interviewer phones each candidate, holds a natural conversation based on your prompt, then hands you back a recording, a transcript, a summary, and a score: no calendars, no scheduling, no recruiter on the line.

1. Add a Voice Call stage

In the pipeline editor, add a Voice Call node and open its settings:

Node Name: how the stage appears in your pipeline.
Preset: start from Screening (Joe) or Screening (Sally) for a ready-made phone screen, or pick Customizable to write your own from scratch.
First Message: the opening line the AI speaks when the candidate picks up.
Prompt: the interview brief, covering what to ask, what to probe, and how to evaluate. Use the Name, Company, Position, and Job Description variables to personalize every call automatically.
Voice: choose Cordial Joe or Laidback Sally.

💡
Not sure how it will sound? Under Try a sample call to test the configuration, enter your own phone number and click Start Sample Call, and the AI will ring you and run the interview exactly as a candidate would hear it.

2. How candidates get the call

There's nothing for the candidate to install or join. When they reach the Voice Call stage, their application page shows a Phone Call Interview step: they enter their phone number and click Start Phone Interview. Their phone rings within moments and the AI interviewer takes it from there. It's a real phone call, not a browser session.

If the call doesn't come through, the candidate can click Didn't get a phone call? Try again after a short cooldown, for up to four attempts in total.

⚠️
Voice interviews consume credits based on call length, so make sure your workspace has balance before activating the stage. See Credits & Billing.

3. Review the results

Once the call ends, Hivemind processes it automatically. Open the candidate's profile and check the Calls tab (the entry is labeled AI voice call), or open their report from the pipeline. You'll find:

Recording: the full audio, playable in the browser or downloadable.
Transcript: the complete conversation, with a Copy button.
Summary: an AI-written recap of how the interview went.
Score: an overall result out of 100.

The AI's evaluation can also drive your pipeline automatically: follow the Voice Call stage with a Flow node to route strong performers forward and screen out the rest. See the Pipeline Nodes Reference.

4. Prefer a human? Use Manual Phone Call

The Manual Phone Call node schedules recorded phone screens that you conduct. It has three tabs:

Main Config: name the stage, pick a phone number, and set your Company's working hours. This node uses your own Twilio account: if you haven't linked one, you'll see Twilio Account Not Connected with a Connect Twilio Account button.
Scheduler: candidates who reach the stage appear here; click Schedule to book a time. Both sides get an email with a calendar invite.
Awaiting Review: after each call, listen back and click Approve or Decline to decide who moves on.

At the scheduled time:

Open the call from your email link.
Enter Your Phone Number.
Click Start Conference Call.

Hivemind dials you and the candidate into the same call, and records and transcribes it just like an AI interview.

What's next

Browse recordings and transcripts across your workspace in the Calls Log
Combine phone screens with tests in Assessments Overview
Book live interviews with Calendar & Scheduling