Internal Sourcing Agent (AI-Powered Database Search)

Last updated: January 9, 2026

Feature Overview

The Internal Sourcing Agent is an AI-powered co-pilot that intelligently searches your entire ATS database using natural language queries. Unlike traditional ATS search that relies on basic keywords, this agent searches every data point for every candidate - including CVs, cover letters, attachments, notes, interaction history, and application forms. It can identify candidates based on skills, experience, behaviors, and attitudes, solving the common problem where 80% of externally sourced candidates already exist in your database.

Available Actions

This feature enables you to perform the following actions:

  • Search database with natural language - Use conversational queries or voice commands to find candidates

  • Search role-specific candidates - Automatically apply job requirements from active analyses

  • Refine searches iteratively - Use AI suggestions to drill deeper into results

  • Search by behaviors and attitudes - Find candidates demonstrating specific traits beyond just skills

  • View match scores and explanations - Understand why each candidate was selected

  • Save and share searches - Collaborate with team members on candidate identification

  • Invite candidates to apply - Send AI-powered personalized outreach directly from results

  • Track outreach campaigns - Monitor delivery, responses, and conversions

Step-by-Step Instructions

Accessing the Search Agent

Two Access Methods:

Method 1: General Database Search

  1. Navigate to Candidates section

  2. Click in the "Ask Popp Anything" chat interface

  3. Type your query naturally

  4. Review results on the left with match scores

Method 2: Role-Specific Search

  1. Navigate to Analyse → Select your role

  2. Click Source More Candidates

  3. System auto-applies job requirements as search parameters

  4. Review pre-filtered results and refine as needed

Conducting Searches

Using Natural Language

Example Queries:

  • "Machine learning engineers in the UK with 2+ years in their current role"

  • "Senior data engineers with AWS and Python who have led teams"

  • "Marketing managers with B2B SaaS experience who have demonstrated entrepreneurial traits"

Understanding Results

Results Display:

  • Candidates appear on left with match scores (0-100%)

  • Higher scores = stronger alignment with criteria

  • AI provides detailed explanation of search interpretation

  • Top candidates include summary of why they matched

Quick Verification:

  • Check location matches your requirement

  • Verify job titles align with search

  • Confirm experience levels are correct

  • Review key skills presence

Refining Searches

AI-Suggested Refinements

After displaying results, AI suggests refinement options as clickable pills:

  • Technical skills (AWS, GCP, Docker)

  • Domain expertise (NLP, Computer Vision)

  • Industry experience (Startup, FinTech)

To Use:

  1. Click any suggested pill to add it to your next query

  2. Or type additional criteria naturally

  3. System combines previous + new requirements

  4. Results update automatically

Iteration Strategy

Start Broad, Then Narrow:

  1. Begin with basic requirements (role, location, core skills)

  2. Review results and match scores

  3. Add specific technical skills or behavioral criteria

  4. Use AI suggestions to guide refinement

When to Tighten:

  • Too many results returned

  • Match scores are lower than desired

  • Need more specific skill combinations

When to Broaden:

  • Few or no results

  • Search was too restrictive

  • Want to explore adjacent skill sets

Searching by Behaviors and Attitudes

What It Enables: Find candidates demonstrating specific traits based on career history and documentation, not just keywords.

Example Behavioral Searches:

  • "Entrepreneurial traits" → Finds founders, startup experience, side projects

  • "Leadership qualities" → Finds management experience, team building, mentoring

  • "Innovation-focused" → Finds patent holders, R&D roles, product development

AI Interpretation: The AI explains how it defines each behavioral trait and what evidence it looks for, ensuring transparency in matching.

Combining Criteria: "ML engineers in UK with AWS skills who have demonstrated entrepreneurial traits" combines technical requirements with behavioral matching for powerful results.

Reviewing Candidate Profiles

Accessing Profiles:

  1. Click any candidate name from results

  2. View comprehensive information including:

    • CVs, cover letters, attachments

    • Contact information and work history

    • Skills (from custom taxonomy specific to your database)

    • Application history within your system

    • Interaction history and notes

    • Conversational campaign participation

Custom Skills Taxonomy:

  • Popp creates a unique skills taxonomy for your database

  • Not generic - tailored to your industry and candidate pool

  • Dramatically improves search accuracy

  • Automatically extracts and categorises skills

Inviting Candidates to Apply

Selection and Invitation

  1. Select candidates using checkboxes

  2. Click Invite to Apply

  3. System opens outreach configuration workflow

Outreach Configuration

Step 1: Review Selection

  • Confirm correct candidates included

  • Verify contact information available

Step 2: Configure Job Information

  • Job type (full-time, contract, etc.)

  • Location and contract details

  • Additional context beyond job description

  • Benefits, culture, growth opportunities

Step 3: Set Campaign Goal Choose desired candidate action:

  • Upload CV (recommended) → Auto-creates application in ATS

  • Schedule Call → Integrates with Scheduling Agent

  • Application URL → Redirects to external system

Step 4: Customize Messaging

Agent Identity:

  • Name your AI agent

  • Customize email subject line

Opening Message:

  • AI generates personalized email body automatically

  • Edit any part of the message

  • Use personalization tags:

    • {First Name} {Last Name}

    • {Job Title} {Organisation Name}

    • {Candidate Relevance} → AI-generated explanation of why they're a good match

Example Candidate Relevance: "I came across your background and was impressed by how you led ML pipeline development at Motorway using Google Cloud and Kubernetes. Your MLOps expertise aligns perfectly with what we're looking for."

Step 5: Preview Individual Messages

  1. Click Review Invitations

  2. View exact message each candidate receives

  3. Edit individual messages if desired

  4. Save changes per candidate

Step 6: Send Invitations

  • Click Send Invitations to launch campaign

  • AI agent takes over all subsequent interactions

From Invitation to Application

Automatic Application Creation

When Candidates Upload CVs:

  1. Candidate attaches CV in response

  2. Application automatically created in ATS

  3. Application also created in Popp

  4. Candidate moves to All Applicants section

Analysis Workflow

Automatic Deep Assessment:

  • Analysis Agent evaluates candidate against role requirements

  • Reviews CV plus all documentation on file

  • Generates detailed suitability report including:

    • Match score and requirement fulfillment

    • Skills assessment

    • Experience evaluation

    • Recommendations for next steps

Activate Additional Workflows:

  • Send to screening campaigns

  • Trigger interview scheduling

  • Direct to assessments

  • Manual review and shortlisting

Saving and Sharing Searches

Search History

  • All queries automatically saved

  • Return to previous searches anytime

  • Continue refining from where you left off

Best Practices

Query Formulation

Effective Queries Include:

  • Job title or role type

  • Key technical skills

  • Experience level/tenure

  • Location requirements

  • Critical qualifications

Use Natural Language:

  • Don't use Boolean operators (AND, OR, NOT)

  • Speak/type conversationally

  • Be specific but not overly complex initially

Example Strong Query: "Senior data engineers with 5+ years experience in London who have worked with large-scale data pipelines and have Python and Spark skills"

Interpreting Results

Match Score Guide:

  • 90%+: Very strong alignment

  • 70-89%: Solid matches worth reviewing

  • 50-69%: Partial matches, may need development

  • <50%: Weak matches, likely not suitable

Few Results? Query too specific—broaden criteria Many Results? Query too broad—add refinements

Leveraging AI

Pay Attention To:

  • How AI interprets your query (verify it matches intent)

  • Suggested refinement pills (faster than typing)

  • Candidate match explanations (understand why they scored highly)

  • AI's definition of behavioral traits (ensure alignment)

Compound Searches

Combine multiple criteria types for powerful results:

  • Technical requirements (skills, tools, platforms)

  • Experience requirements (years, seniority)

  • Behavioral traits (entrepreneurial, leadership)

  • Industry knowledge (FinTech, SaaS)

  • Geographic preferences (location, remote)

Example: "Senior backend engineers with 7+ years, UK-based, Python and Go, AWS architecture, who have demonstrated leadership through mentoring, preferably with FinTech background"

Troubleshooting Common Issues

No or Few Results

Solutions:

  • Remove some optional requirements

  • Broaden geographic area

  • Reduce experience minimums

  • Check spelling of technical terms

  • Use more general job titles

Results Don't Match Expectations

Check:

  • AI's interpretation explanation

  • Expanded titles or skills match your intent

  • Behavioral criteria defined as expected

Adjust:

  • Rephrase ambiguous terms

  • Add clarifying requirements

  • Use different terminology

Outreach Not Sending

Verify:

  • Valid email addresses in candidate records

  • Check delivery status in dashboard

  • Review bounce notifications

  • Ensure candidates haven't unsubscribed

Quick Reference

Access:

  • General: Candidates → "Ask Popp Anything"

  • Role-Specific: Analyse → Role → Source More Candidates

Search Capabilities:

  • Every data point searchable (CVs, notes, attachments, history)

  • Behavioral and attitudinal matching

  • Custom skills taxonomy

  • Voice and text queries

  • Iterative refinement with AI suggestions

Outreach Setup:

  • Configure job details and campaign goal

  • Customize agent and messaging

  • Preview individual messages

  • AI handles all follow-up automatically

Integration:

  • Automatic application creation in ATS

  • Flow to Analysis Agent for evaluation

  • Connect to screening campaigns

  • Link to interview scheduling

Key Benefits:

  • Eliminates 80% duplication with external sourcing

  • Searches millions of candidates instantly

  • Finds candidates based on behaviors, not just keywords

  • Seamless workflow from search to application

  • AI-powered personalized outreach at scale