urlcv
Back to Documentation

AI Relationship Graph (Beta)

📋 On This Page

  • Overview
  • What This Feature Does
  • Why It Matters
  • How It Works
  • Step-by-Step Workflow
  • Outputs
  • Best Practices
  • Limitations
  • Related Features

    • ---

    Overview

    The AI Relationship Graph automatically builds a relationship network from candidate work histories, identifying who has worked together, at which companies, for how long, and when. This feature helps recruiters surface hidden networks, identify warm introductions, find referral paths, and add context to candidate submissions.

    ---

    What This Feature Does

    The AI Relationship Graph analyzes work history data from parsed CVs to create a network of professional relationships based on shared employment. When candidates have overlapping employment periods at the same company, the system calculates relationship strength, overlap duration, and provides relationship context that recruiters can use to:

  • Surface warm introductions — Find candidates who worked together and can provide references
  • Identify referral paths — Discover connections between candidates and target companies
  • Add submission context — Show clients which candidates have worked together before
  • Find company alumni — Search all candidates who worked at a specific company during a date range
  • Strengthen placements — Highlight team chemistry and pre-existing working relationships

    • ---

    Why It Matters for Agencies

    Faster Shortlisting with Relationship Context

    When you know two candidates worked together for 18 months at the same company, you can confidently present them as a team or use one to reference the other, reducing client risk and increasing placement velocity.

    Unlock Hidden Referral Networks

    Search for all candidates who worked at a target company during a specific period to find warm introductions, backdoor references, and relationship paths that cold outreach can't access.

    Improve Placement Quality

    Highlight candidates who've successfully worked together before, reducing integration risk for clients and demonstrating deep market knowledge that differentiates your agency.

    Compliance-Safe Relationship Tracking

    All relationship data derives from work history already stored in candidate profiles, with full GDPR audit trail coverage and no external data scraping.

    ---

    How It Works

    Work History Extraction

    When CVs are parsed using AI CV Parsing, work history entries are extracted and stored with:

  • Company name
  • Job title
  • Start date
  • End date (or "Present" for current roles)
  • Location
  • Description

    • Company Name Normalization

    The system normalizes company names to match variations:

  • Strips common suffixes — Ltd, Inc, Corp, Limited, LLC, Co, PLC
  • Removes special characters — Punctuation and symbols
  • Lowercases text — Case-insensitive matching
  • Trims whitespace — Removes extra spaces

    • Example:

  • "Acme Corporation Ltd." → "acme"
  • "ACME Inc" → "acme"
  • "Acme Co." → "acme"

    • All three normalize to the same company for matching purposes.

    Overlap Detection

    The system calculates temporal overlap between two candidates' work periods:

  • Find overlap window — Identify the start and end of the overlapping period
  • Calculate duration — Measure overlap in days, months, and years
  • Calculate percentage — Overlap as percentage of shorter tenure
  • Store relationship — Record connection with all context

    • Overlap Calculation:

    Overlap Start = MAX(Candidate1 Start, Candidate2 Start)
Overlap End = MIN(Candidate1 End, Candidate2 End)
Overlap Duration = Overlap End - Overlap Start
Overlap Percentage = (Overlap Duration / Shorter Tenure) × 100

    Relationship Confidence Scoring

    Relationship strength is indicated by:

  • Overlap duration — Longer overlaps = stronger relationships
  • Overlap percentage — Higher percentage = more significant connection
  • Job titles — Similar seniority levels may indicate peer relationships
  • Time period — Recent overlaps may indicate active relationships

    • ---

    Step-by-Step Workflow

    1. Upload CVs and Parse Work History

  • Navigate to Candidates → Upload CV
  • Upload candidate CVs (PDF format)
  • AI extracts work history automatically
  • Work history appears on candidate profile
  • Repeat for multiple candidates

    • Note: Work history must be present in CVs for relationship detection to work.

    2. View Candidate Connections

    From Candidate Profile:

  • Open any candidate profile
  • Scroll to Work History section
  • Timeline view shows all employment periods
  • Click View Connections to see relationships
  • System displays all candidates with overlapping work periods

    • Connection Details Shown:

  • Connected candidate name and current role
  • Shared company name
  • Both candidates' job titles during overlap
  • Overlap period (start and end dates)
  • Overlap duration (months/years)
  • Overlap percentage
  • Links to both candidate profiles

    • 3. Search by Company

    Company Search:

  • Navigate to Work History → Company Search
  • Enter company name (variations will match due to normalization)
  • (Optional) Set date range to filter by time period
  • System shows all candidates who worked there
  • Filter results by:
    • - Location — Office location during employment - Role category — Job function or seniority - Date range — Time period at company - Placement status — Placed at company, placed anywhere, or all candidates

    Use Cases:

  • Find all candidates who worked at Target Company between 2018-2020
  • Identify warm introductions to hiring managers at a client
  • Search for alumni from acquired companies
  • Find candidates who overlapped with a specific hire

    • 4. View Client Company Connections

    From Client Company Profile:

  • Navigate to Client Companies → Select company
  • Click View Connections tab
  • System shows all candidates with work history at that company
  • See candidate names, roles, employment periods, and duration
  • Use for:
    • - Finding warm introductions to client - Identifying candidates with company-specific experience - Showing relationship depth in pitches

    5. Use Relationships in Submissions

    Adding Relationship Context:

    When creating submissions, reference relationship insights:

  • "Candidate A and Candidate B worked together at Acme Corp for 18 months (2019-2020)"
  • "Candidate has 3 years experience at Target Company (2015-2018), overlapping with current hiring manager"
  • "We have 12 candidates with experience at Competitor Co during this period"

    • Submission Notes:

    Add relationship context to submission notes to:

  • Highlight team chemistry
  • Show backdoor references
  • Demonstrate market knowledge
  • Reduce client perceived risk

    • ---

    Outputs

    Relationship Objects

    Each detected relationship includes:

    Candidate Information:

  • Candidate A (source candidate)
  • Candidate B (connected candidate)
  • Both candidates' current roles

    • Company Context:

  • Shared company name (original and normalized)
  • Company location (if available)

    • Time Period:

  • Candidate A's employment period at company
  • Candidate B's employment period at company
  • Overlap start date
  • Overlap end date
  • Overlap duration (days, months, years)
  • Overlap percentage (relative to shorter tenure)

    • Job Context:

  • Candidate A's job title during overlap
  • Candidate B's job title during overlap
  • Role categories (if available)

    • Company Search Results

    Candidate List:

  • All candidates who worked at searched company
  • Employment periods and job titles
  • Duration at company
  • Overlap with specified date range (if provided)
  • Current candidate status

    • Available Filters:

  • Locations — Where candidates worked (with counts)
  • Roles — Role categories (with counts)
  • Date Range — Earliest and latest employment dates across all candidates
  • Placement Status — Placed at company, placed anywhere, all candidates

    • ---

    Best Practices

    1. Ensure Complete Work History

    During CV Upload:

  • Verify AI extracted all employment periods
  • Add missing companies manually if needed
  • Include start and end dates for accurate overlap calculation
  • Update "Present" for current roles

    • Data Quality:

  • More complete work history = more relationship insights
  • Missing dates reduce overlap detection accuracy
  • Consistent company names improve matching

    • 2. Use Company Search Strategically

    Target Company Intelligence:

  • Search client companies before pitches to find warm introductions
  • Search competitor companies to find industry specialists
  • Search recently acquired companies to identify talent pools

    • Date Range Filtering:

  • Narrow to recent periods (last 2-5 years) for active relationships
  • Search historical periods for long-term industry veterans
  • Match client's company timeline for cultural fit

    • 3. Add Relationship Context to Submissions

    Strengthen Submissions:

  • Reference shared employment in submission notes
  • Highlight team chemistry and proven collaboration
  • Show backdoor reference availability
  • Demonstrate deep market knowledge

    • Client Communication:

  • "These two candidates worked together at Acme for 18 months"
  • "Candidate has experience at your competitor during high-growth period"
  • "We have 8 candidates from Target Company's engineering team"

    • 4. Verify Relationships

    Before Using in Submissions:

  • Check overlap duration is significant (3+ months recommended)
  • Verify job titles suggest actual collaboration (same department/team)
  • Consider company size (10-person startup vs 10,000-person corp)
  • Ask candidates about relationship if using for references

    • Context Matters:

  • 6-month overlap at 20-person startup = strong relationship likely
  • 6-month overlap at 5,000-person corporation = may not have interacted
  • Same job title + overlap = peer relationship likely
  • Different seniority + overlap = may not have worked together

    • 5. Maintain Data Privacy

    GDPR Compliance:

  • Work history data covered by existing audit trail
  • Relationship insights derived from candidate-provided CVs
  • No external data scraping or enrichment
  • Candidates control work history through profile updates

    • Use Anonymisation:

  • Anonymise candidates when sharing relationship insights
  • Don't reveal candidate identity without consent
  • Use relationship context only after NDA acceptance

    • ---

    Limitations (Beta)

    Current Limitations

    Company Matching:

  • Relies on text normalization, not company entity resolution
  • Different divisions or subsidiaries may not match
  • Acquisitions and name changes require manual mapping
  • International variations may not normalize correctly

    • Overlap Detection:

  • Only detects direct employment period overlaps
  • Doesn't infer team membership or reporting relationships
  • Can't distinguish between different office locations at same company
  • No detection of project-based collaboration

    • Relationship Strength:

  • Duration-based only, no qualitative assessment
  • Can't determine if candidates actually worked together
  • No sentiment or relationship quality scoring
  • Overlap percentage helps but isn't definitive

    • Data Requirements:

  • Requires parsed work history (from CV upload or manual entry)
  • Missing dates reduce accuracy
  • Incomplete work history limits relationship discovery
  • Historical CVs may omit early career details

    • Future Enhancements

    Planned Improvements:

  • AI-inferred relationship strength based on job titles and company size
  • Automated relationship summaries using GPT-4o
  • Company entity resolution with external data sources
  • Team and project-based relationship detection
  • Relationship confidence scores
  • Social graph visualization
  • Referral path discovery (multi-hop connections)

    • ---

    🔗 Related Features

  • AI CV Parsing — Work history extraction from CVs
  • Candidate Management — Maintaining candidate work history
  • AI Candidate Matching — Enhanced matching with relationship context
  • Submissions & Tracking — Using relationship insights in submissions
  • GDPR Audit Trail — Compliance coverage for relationship data