Use Claude to screen candidates before human interviews

micro1 used Claude to run live technical screens and produce reports before handing qualified candidates to human interviewers.

Done with ClaudeΒ· Advanced

What they did
micro1 built its recruiting interview flow on the Claude Platform API. Claude chatted with candidates in real time, at any hour and across time zones, and then generated an interview report for the hiring team. People who cleared that first pass moved on to human interviewers.
How it went
The vendor case study says the system handled more than 3,000 interviews a day and cut recruiting costs by 85%. It also says Deel’s human interview pass rate increased from 10% to 50%, while Legal Soft reduced its screening staff from 33 to 12. Those figures are self-reported and the case study does not explain its scoring or methodology.
Tips
If you try this, keep a human in charge of final hiring decisions and review how the model grades candidates. Make sure the interview flow is tested for consistency before using it at scale.

Try it yourself with Claude

You are screening candidates for a technical role at [company]. Interview me one question at a time about [skills or topics], and adjust your follow-up based on my answers. If I give a vague or incomplete answer, ask for clarification or a concrete example. After [number] questions, pause and show me a draft scorecard with a 1–5 rating for each skill, brief evidence from my responses, and a hire / no-hire recommendation. Wait for my approval before you send, submit, or use that recommendation anywhere.

Fill in the [brackets]. Agents can make mistakes, so review anything before it is sent, paid or deleted.

Read the original on anthropic.com β†—

Get 3 agent uses like this every morning

With the day's top AI news. Free, and one click to unsubscribe.