Professional Services9 min read

Executive Search with Claude: PDPA-Safe CV Parsing at Scale

How Singapore headhunters use Claude to parse, score, and shortlist 500 CVs in under an hour, without breaching PDPA. Full prompt and workflow.

Haojun See

Founder & Director, On The Ground

Updated 1 May 2026

The headhunter's bottleneck

For most Singapore executive-search engagements, the bottleneck isn't sourcing — it's processing the long list down to a credible shortlist of 5–8 candidates the client meets. 200–500 CVs / LinkedIn profiles per role isn't unusual. Manual processing: 8–15 hours of senior consultant time. That time is the headhunter's biggest cost line. Claude compresses this 80%, with arguably better consistency than a tired senior consultant on CV #287.

The CV-parse-and-score prompt

*"You are screening a candidate for [ROLE TITLE] at a [CLIENT TYPE — e.g. SG-based logistics SME, ~150 staff]. Search criteria from the engagement letter: - Must-haves: [LIST] - Strong nice-to-haves: [LIST] - Red flags: [LIST] - Salary range: [BAND] Below is a candidate CV. Score them and return JSON: - name: REDACTED - years_relevant_experience: number - must_haves_met: array - must_haves_missing: array - strong_nice_to_haves_met: array - red_flags_present: array - top_3_strengths: array (specific to this role) - top_3_concerns: array (specific to this role) - comp_signals: any mentions of compensation, equity, notice period - one_sentence_summary - recommend_for_initial_call: yes / no / maybe Use Singapore English. Be specific — avoid generic phrases. CV: [PASTE]"* Run programmatically across the long list. Sort by recommend_for_initial_call + must_haves_met count. Senior consultant reviews top 30, picks shortlist of 8.

Identifier redaction before scoring

Critical step. Before passing CV text to the scorer, run a redaction pre-pass: *"Below is a CV. Produce a redacted version that removes: full name (replace with [CANDIDATE]), age / DOB, schools attended (replace with 'top SG university' / 'overseas university' / 'polytechnic' / 'other' as appropriate), specific company names that could identify the candidate via Google (replace with industry+size+role description). Preserve: years of experience, role titles, achievements, technical skills, salary information, notice period. Output only the redacted CV. CV: [PASTE]"* The scoring prompt then runs against the redacted version. Identifiers reach the headhunter at the human-review stage — *after* a fit-score has been assigned blind. This isn't perfect bias mitigation, but it materially reduces the risk of the AI surfacing schools, age, or company names as hidden ranking factors.

Generating the candidate brief

For shortlisted candidates, Claude can draft the candidate brief that goes to the client: *"Below is a candidate CV (de-identified) and our notes from the screening call. Write a 1-page candidate brief in our standard format: - Role fit (2 sentences) - Career narrative (3 bullets — what shaped them, key achievements, current chapter) - Strengths for this role (3 bullets) - Risks / development areas (2 bullets, candid but not damaging) - Why this opportunity fits them (2 sentences) - Compensation expectations - Notice period Tone: factual, partner-level, Singapore commercial English. No clichés. CV: [PASTE] Call notes: [PASTE]"* Senior consultant reviews and edits — 15 minutes vs. 60 minutes from scratch.

The pipeline as a tool

For continuous use (a recruitment business doing this every week), build it as a tool: 1. Drop CVs in a folder (PDFs, Word, .txt). 2. Auto-parse text, redact identifiers, score against JD. 3. Spreadsheet output sorted by fit_score. 4. One-click candidate brief generation. 5. PDPA-compliant logging and 6-month retention. Total build time as a Functional App Sprint: 5–7 days. Cost: S$3,500–S$5,500. Time saved per engagement: 8–12 hours of senior consultant time. Most SG-based search firms doing 30+ engagements a year have a payback inside Q1.

Where to next

For broader SME hiring, see AI Hiring for SMEs: From JD to Shortlist with Claude — many of the prompts overlap. For PDPA framework details, see PDPA-Safe Claude Prompts: A Lawyer's Checklist (most rules translate to recruitment) and PDPA Prompting Checklist. OTG has built CV-screening pipelines for SG-based search firms and HR teams. Book a free 30-minute call to discuss.

Frequently asked questions

Is it PDPA-safe to upload a candidate's CV to Claude?

Yes when (a) using enterprise-tier Claude (no training on input), (b) the candidate has provided their CV to be processed for placement, and (c) you have appropriate retention policies. Treat Claude as a data processor and document accordingly. See [PDPA Prompting Checklist](/resources/pdpa-prompting-checklist).

Won't AI-assisted shortlisting introduce bias?

AI mirrors the patterns in your prompt. Mitigate: redact identifiers (name, age, school) before scoring, score on JD-relevant criteria only, and have a human review every shortlist. The [Model AI Governance Framework for Generative AI](https://aiverifyfoundation.sg/wp-content/uploads/2024/05/Model-AI-Governance-Framework-for-Generative-AI-May-2024-1-1.pdf) is the Singapore-relevant guidance.

Should I tell candidates AI is involved?

Yes — your privacy notice should disclose AI-assisted processing. Some firms add a one-liner to their CV-submission portal: 'Initial CV review is AI-assisted; all shortlists are human-reviewed before contact.'

How does this compare to specialised tools like HireVue or Phenom?

Specialised ATS / recruitment tools have rich integrations with job boards and pipeline management. For mid-tier SG firms doing executive search, a Claude-based custom workflow is 10× cheaper, more flexible, and equally capable for the parse-and-score step.

Does it work for senior C-suite searches where CVs are scarce?

Yes — the workflow scales from 10 CVs to 500. For very senior searches where CVs are scarce, Claude's value shifts toward synthesising LinkedIn profiles, public references, and earlier career patterns into a candidate brief.

PreviousAI Slide Generation for Consultants: Claude Frameworks That Actually Work Next Claude for CFOs: From Raw Ledger to Board Pack

Want to Apply This to Your Business?

We're a Singapore AI development and automation agency. Let's discuss how we can help solve your specific challenges.

Get in Touch View our services