Use Case
AI CV Screening with Claude Cowork
Stop spending Mondays buried in CVs. Claude scores candidates against your role criteria in seconds — consistently, with a clear rationale you can defend.
30 min
→
2 min
to screen one CV properly
The Problem
For an in-house recruiter or agency, screening CVs is the most time-expensive part of the role. A senior consultant takes 20-30 minutes to do it properly — reading the CV, comparing to the spec, weighing experience against red flags, deciding whether the candidate goes to the longlist. Multiply that by 50 CVs per role, and a single vacancy eats two full working days before anyone's been on the phone.
The shortcut most teams take is keyword scanning. Quick, but unreliable — you miss great candidates who don't use the exact buzzwords, and you pass through weak ones who learned how to game the system. The result: poor first-stage shortlists that waste the next round of time.
Claude Cowork sits in the middle. It does the structured first-pass screening — reading every CV with the same level of attention as your best recruiter on a good day — and produces a scored, justified summary. You spend the saved time on the part of the job AI can't do: the calls, the relationships, the judgement.
What Claude Does Here
- Extracts key information from any CV format in seconds
- Scores candidates against role-specific criteria you define
- Writes a 3-line summary explaining the score
- Flags red flags (gaps, frequent moves, mismatched experience)
- Surfaces strengths a keyword scan would miss
- Generates candidate briefings ready to send to clients
How to Set It Up
1
Define the role criteria explicitly
List the must-haves, the nice-to-haves, and the deal-breakers. Be specific. "5+ years recruitment experience" not "experienced recruiter." Claude scores against exactly what you specify.
2
Build a screening prompt template
Combine the role criteria + your scoring scale (e.g. 1-10) + the output format you want (summary, score, recommendation). Save this as your reusable prompt for the role.
3
Add bias-prevention instructions
Explicitly tell Claude to ignore name, age, university name, and address. Score only on experience, skills, and demonstrated outcomes. This is a 30-second addition to your prompt.
4
Paste each CV into the prompt
One CV at a time. Claude returns a scored summary in 30-60 seconds. You can also batch CVs into a single prompt if you want a comparative ranking.
5
Make the human call on the shortlist
Claude does the structured screening. You make the final shortlist decision — especially for borderline candidates where context matters. This is the GDPR-compliant approach.
Example Prompt
Prompt
Screen this CV against the role criteria below. Score 1-10 and provide a 3-line summary justifying the score.
ROLE: Senior Recruitment Consultant, financial services, London-based agency
MUST-HAVES:
- 4+ years recruitment experience (agency-side)
- Demonstrable billings of £200k+ in last 12 months
- Experience placing into asset management or investment banking
NICE-TO-HAVES:
- Existing client relationships in financial services
- Experience training junior consultants
DEAL-BREAKERS:
- More than 3 job moves in last 5 years
- No agency experience (in-house only)
OUTPUT FORMAT:
Score: [1-10]
Verdict: [Yes / Maybe / No]
Summary: [3 lines max explaining the score]
Strengths: [bullet list]
Concerns: [bullet list]
Suggested next step: [phone screen / longlist / reject]
IMPORTANT: Do not consider name, age, university prestige, or address in your scoring. Score only on experience, demonstrated outcomes, and stated skills.
[paste CV here]
What Claude Returns
Claude's Output
Score: 8
Verdict: Yes
Summary: Strong candidate with 6 years of agency-side recruitment in financial services. Billings of £340k last year (well above threshold), with stated focus on asset management. Two job moves in 5 years — explained by progression, not red flag.
Strengths:
- 6 years agency experience (exceeds 4-year requirement)
- £340k billings in last 12 months
- Asset management specialism explicitly stated
- Promoted internally before each move
Concerns:
- Most recent role only 11 months — worth probing the move
- No mention of training/mentoring junior consultants
Suggested next step: Phone screen. Lead with questions about the most recent move and whether they've done any team development work. Strong fit on the must-haves — just need to confirm stability and explore the nice-to-have on training.
★ Free 5-Day Challenge
Try this for yourself in 10 minutes
The free 5-Day Claude Challenge teaches you exactly how to set up workflows like this — emails, reports, data extraction, and your own automation playbook. No card, no fluff.
Frequently Asked Questions
Is using AI for CV screening compliant with UK GDPR?
Yes, when done correctly. The key requirements: a human makes the final decision (not the AI), candidates are informed AI is used in screening, and you don't store CV data in Claude beyond the active conversation. Our course covers GDPR-compliant setup for recruitment workflows in detail.
Won't AI screening introduce bias?
AI can introduce bias if you let it — but used carefully, it actually reduces it. Tell Claude to score CVs only against specific job criteria you define (skills, experience, qualifications) and explicitly instruct it to ignore name, age, university prestige, and other proxies for protected characteristics. The result is more consistent than human screening.
Can Claude write candidate briefings from screened CVs?
Yes — and this is where the time savings really compound. Once Claude has screened a CV against your criteria, it can immediately produce a candidate briefing for your client in the same conversation. Screening + briefing in under 5 minutes per candidate.
What if the CV is in a weird format like PDF or scanned image?
Claude handles PDFs natively. For scanned image CVs, you'll need to OCR them first (Adobe, Apple Notes, or Google Drive can all do this in 30 seconds). Once the text is extracted, Claude treats it like any other CV.
How accurate is Claude's screening compared to a recruiter?
For first-pass screening against defined criteria, accuracy is comparable to an experienced recruiter — and significantly more consistent. Claude doesn't get tired on CV #47 like a human does. The recruiter's expertise is still essential for the final shortlist call, where context and gut feel matter.
Related Use Cases