Using Smart Agents

Learn how to use Smart Agents to extract any custom information from LinkedIn profiles using AI. This guide provides practical examples and best practices.

Quick Start

1. Open any lead list
2. Click "Add Smart Agent"
3. Name your column (e.g., "Years of Experience")
4. Write your query (e.g., "How many total years of work experience?")
5. Select leads to process (all, first 10, first 20, or custom)
6. Click "Create" and watch results appear

Query Examples by Use Case

Sales Qualification

Decision Maker Identification

Query: "Does this person have budget authority or decision-making power? Return Yes/No"
Expected Output: Yes/No
Use: Filter for decision makers before outreach

Company Size Verification

Query: "What is the employee count at their current company? Return as a number"
Expected Output: 1200
Use: Segment by company size for targeted messaging

Tech Stack Discovery

Query: "What CRM system does this person mention using? Return the CRM name or 'None'"
Expected Output: Salesforce, HubSpot, None
Use: Identify prospects using competitor products

Recruiting

Experience Level

Query: "How many years of software engineering experience does this person have? Return only the number"
Expected Output: 8
Use: Filter candidates by experience level

Specific Skill Check

Query: "Does this person have experience with React and Node.js? Return Yes/No"
Expected Output: Yes/No
Use: Technical skill matching

Management Experience

Query: "Has this person managed a team? If yes, return the size of the largest team managed"
Expected Output: "Yes - 12 people" or "No"
Use: Leadership role qualification

Market Research

Industry Experience

Query: "How many years has this person worked in healthcare/medical industry?"
Expected Output: 6
Use: Industry expertise validation

Geographic Mobility

Query: "List all cities where this person has worked. Return as comma-separated list"
Expected Output: "San Francisco, New York, Austin"
Use: Assess relocation likelihood

Career Progression

Query: "What is this person's career trajectory? Return: Ascending, Lateral, or Descending"
Expected Output: Ascending
Use: Identify high-performers

Writing Effective Queries

Be Specific

Vague: "Tell me about their experience" Specific: "How many years of B2B SaaS sales experience does this person have?"

Vague: "Do they know Python?" Specific: "Does this person list Python in their skills or mention using it in their work experience? Return Yes/No"

Specify Output Format

For Yes/No questions:

"Does this person work in enterprise sales? Return Yes/No"

For Numbers:

"How many companies has this person worked for? Return as a number only"

For Lists:

"What programming languages does this person know? Return as comma-separated list"

For Text (keep concise):

"What is this person's current job title? Return title only, max 50 characters"

Ask One Thing Per Column

Multiple questions: "What's their experience and do they have budget authority and what tech do they use?"

Single focused query: "Does this person have budget authority? Return Yes/No" (Create separate columns for other questions)

Processing Strategies

Test First (Recommended)

1. Select "First 10" leads
2. Review results for accuracy
3. Refine query if needed
4. Process remaining leads
5. Saves credits on bad queries

Selective Processing

For large lists (1000+ leads):

1. Sort or filter first (e.g., by company size, title)
2. Select top N leads that match criteria
3. Process only qualified subset
4. Export and use elsewhere

Batch Processing

For multiple Smart Agents:

1. Create all columns first (don't process yet)
2. Select same lead subset for all
3. Process simultaneously
4. More efficient use of API calls

Real-World Examples

Example 1: ABM Campaign Qualification

Objective: Find VP+ level contacts at companies using Salesforce

Setup:

• Column 1: "Is this person VP-level or above? Return Yes/No"
  • Process: All leads
• Column 2: "Does this person mention Salesforce in their profile? Return Yes/No"
  • Process: All leads

Filter: Both columns = "Yes" Result: 47 qualified leads from 500-lead list Action: Export to sales team for targeted outreach

Example 2: Engineering Recruitment

Objective: Find senior backend engineers with cloud experience

Setup:

• Column 1: "How many years of backend development experience?"
  • Process: All leads
• Column 2: "Does this person have AWS or Google Cloud experience? Return which one or Both or Neither"
  • Process: All leads
• Column 3: "Has this person worked at a company with 1000+ employees? Return Yes/No"
  • Process: Leads with 5+ years experience

Filter: Column 1 >= 5, Column 2 != "Neither" Result: 23 qualified candidates Action: Send to recruiting team

Example 3: Event Targeting

Objective: Find CTOs/VPs in fintech for conference invites

Setup:

• Column 1: "Is this person a CTO, VP Engineering, or Head of Engineering? Return the title or No"
  • Process: All leads
• Column 2: "How many years has this person worked in financial technology/fintech?"
  • Process: Only tech leadership (Column 1 != "No")

Filter: Column 1 != "No" AND Column 2 >= 2 Result: 31 qualified prospects Action: Send personalized conference invitations

Combining with Standard Fields

Smart Agents work great with standard enrichment:

Workflow:

1. Enrich for email/phone first
2. Add Smart Agents for qualification
3. Filter by Smart Agent criteria
4. Export leads with verified contact info
5. Import to CRM for outreach

Example:

Standard Fields: Email, Phone, Company, Title
Smart Agent: "Budget Authority? Yes/No"

Filter: Budget Authority = Yes + Email != empty
Export: 127 contacts ready for outreach

Performance Tips

Query Complexity

Fast queries (3-5 sec/lead):

• Simple yes/no questions
• Extracting visible fields
• Counting work history

Slow queries (8-12 sec/lead):

• Complex analysis
• Multiple conditions
• Inferring implicit information

Credit Usage

Smart Agent pricing varies by complexity:

Simple queries: 1-2 credits

• "What is current job title?"
• "How many years of experience?"

Medium queries: 3-4 credits

• "Does this person have budget authority?"
• "What technologies do they use?"

Complex queries: 5+ credits

• Multi-step analysis
• Inference-heavy questions
• Long-form responses

Optimization

1. Use existing data when possible:

   • Check if standard fields have what you need
   • Review existing Smart Agents
   • Don't duplicate queries

2. Process selectively:

   • Test on subset first
   • Only process qualified leads
   • Use filters before Smart Agents

3. Reuse Smart Agent queries:

   • Save successful queries
   • Apply to new lists
   • Build query library

Troubleshooting

Low Accuracy

Problem: Results don't match expectations

Solutions:

• Make query more specific
• Add examples in query ("e.g., VP Sales, Director Sales")
• Verify information exists in profiles
• Try simpler, more direct query

"Not Found" Results

Problem: Many leads return empty/null

Reasons:

• Information not in LinkedIn profile
• Profile is private/restricted
• Query too specific/narrow
• Asking for information LinkedIn doesn't have

Fix: Adjust query to ask for available information

Inconsistent Format

Problem: Answers in different formats

Solution: Be very explicit about format

"How many years of experience?"

• Returns: "8 years", "8", "Eight", "~8 years"

"How many years of experience? Return only the number with no text"

• Returns: "8", "8", "8", "8"

Advanced Techniques

Conditional Queries

"If this person is in sales, return their quota attainment if mentioned. If not in sales, return 'N/A'"

Multi-Step Analysis

"First check if this person has led a team. If yes, return the size of the largest team. If no, return 'Individual Contributor'"

Confidence Scoring

"Does this person likely have purchasing authority? Return: High Confidence, Medium Confidence, Low Confidence, or No"

Learn More

• Learn about managing lists
• Export your Smart Agent data
• Understand credit pricing
• View Smart Agents documentation
• Smart Agents product page (opens in a new tab)