AI Research Assistant

AI Research lets you automatically gather custom insights about each prospect in your list. These insights can then be used to personalize your AI SDR messages.

What is AI Research?

AI Research is a tool that searches the web and analyzes information to provide structured data based on your requirements. It combines web search with intelligent analysis to extract exactly the information you need.

Think of it as a research assistant that can:

• Search the internet for specific information
• Read and analyze web content
• Extract key data points
• Present findings in a structured format you define

How AI Research Works

Every AI Research request has two parts:

1. Prompt: Your instructions for what to research and how to analyze it
2. Output Settings: The specific fields you want returned

Prompt Structure

Organize your prompt with clear sections:

#CONTEXT#
You are a business research assistant.

#OBJECTIVE#
Find recent news about {{org_company_name}}.

#INSTRUCTIONS#
1. Search for news from the past 6 months
2. Focus on funding, partnerships, or product launches
3. Use reliable sources like news sites and press releases
4. Provide the article title, URL, and a brief summary

Getting Started: Your First Prompt

If you've never written an AI prompt before, use this approach:

Step 1: Think Like You're Talking to a Research Assistant

Instead of thinking about "AI prompts," imagine you're giving instructions to a person.

Example: "I need you to find information about this company's recent news"

Step 2: Be Specific About What You Want

The more specific you are, the better results you'll get.

Vague: "Find company information"
Better: "Find recent news articles about this company from the past 6 months"
Best: "Search for recent funding announcements, partnerships, or product launches for this company from the past 6 months using reliable news sources"

Step 3: Structure Your Request

Use clear sections to organize your instructions:

#CONTEXT#
You are a business research assistant.

#OBJECTIVE#
Find recent news about the company {{org_company_name}}.

#INSTRUCTIONS#
1. Search for news articles from the past 6 months
2. Focus on funding, partnerships, or product launches
3. Use reliable sources like news websites and press releases
4. Provide the article title, URL, and a brief summary

Writing Effective Prompts

Follow these best practices to get the most out of AI Research:

1. Use Clear Structure

Organize your prompt with clear sections:

• #CONTEXT#: What role should the AI take?
• #OBJECTIVE#: What is the main goal?
• #INSTRUCTIONS#: Step-by-step what to do
• #EXAMPLES#: Show expected input/output format (optional)

2. Specify Reliable Sources

Always specify what types of sources are acceptable:

• News articles
• Company websites
• Press releases
• Industry reports
• Government databases
• Specific domains (e.g., crunchbase.com, techcrunch.com)

3. Set Time Constraints

When relevant, specify time ranges:

• "from the last 12 months"
• "recent announcements"
• "within the past 2 years"

4. Use Available Variables

You can reference variables using the {{variable_name}} syntax.

Important Tips:

• Company research: Use both {{org_company_name}} and {{org_domain}} together to reduce ambiguity (unless the company is very well known).
• Personal research: Use personal variables together with {{org_company_name}} and {{org_domain}} to identify the right person.

Available Variables

Use these variables in your prompts to dynamically insert prospect data. Variables should be wrapped in double curly braces when used in prompts (e.g., {{per_first_name}}).

Person Fields

Use these variables to reference individual prospect information:

Organization Fields

Use these variables to reference company information:

What AI Research Can Do

Web Search and Analysis

• Search for recent news and announcements
• Find company information from websites
• Research industry trends and competitors
• Analyze funding rounds and partnerships
• Extract specific data from web sources
• Cross-reference information from multiple sources

Data Extraction and Structuring

• Convert unstructured web content into structured data
• Extract specific fields from complex information
• Categorize and classify information
• Provide confidence scores and reasoning
• Summarize findings in specified formats

Company and Industry Research

• Identify business models (B2B, B2C, etc.)
• Find competitors and market positioning
• Research revenue models and pricing strategies
• Analyze company technologies and services
• Track funding history and partnerships

What AI Research Cannot Do

Real-Time or Live Data

• Cannot access real-time stock prices or live feeds
• Cannot provide up-to-the-minute social media updates
• Cannot access private or restricted databases

Personal or Private Information

• Cannot access private social media profiles (including most LinkedIn profiles)
• Cannot retrieve confidential business information
• Cannot access password-protected content

Complex Calculations or Modeling

• Cannot perform complex financial modeling
• Cannot run statistical analyses on large datasets
• Cannot generate predictive models

Content Creation Beyond Research

• Cannot create marketing materials or sales copy
• Cannot generate images or multimedia content
• Cannot write code or technical documentation

Setting Up Output Settings

Instead of writing complex data structures, you'll use the Output Settings interface.

Field Types Available

Use Output Settings to define the structured fields you want back. Field types include:

• Text: Names, descriptions, URLs, summaries
• Number: Scores, counts, percentages
• True/False: Yes/no questions

Best Practices for Output Settings

1. Use descriptive field names (e.g., company_name instead of name)
2. Choose the right type (Number for scores, True/False for yes/no)
3. Describe what information belongs in each field
4. Plan for missing data or unavailable sources

Example Output Settings

• Field Name: company_name, Type: Text, Description: Official company name
• Field Name: is_saas_company, Type: True/False, Description: Whether the company operates a SaaS business model
• Field Name: confidence_score, Type: Number, Description: Confidence level from 0-100 for the classification

Prompt Examples

Example 1: Company Research

#CONTEXT#
You are a business analyst researching companies for potential partnerships.

#OBJECTIVE#
Research the company {{org_company_name}} with domain {{org_domain}} to determine their business model and recent developments.

#INSTRUCTIONS#
1. Visit the company website {{org_domain}} to understand their services
2. Search for recent news articles about the company from the past 12 months
3. Identify their primary business model (B2B, B2C, or both)
4. Look for any recent funding announcements or partnerships
5. Provide a confidence score for your business model assessment

#EXAMPLES#
Input: Company with e-commerce platform
Output:
- Business Model: "B2C"
- Recent News: "Raised $5M Series A in March 2024"
- Confidence: 95

Example 2: Competitive Analysis

#CONTEXT#
You are conducting competitive intelligence research.

#OBJECTIVE#
Identify the top 3 direct competitors of {{org_company_name}} based on their website {{org_domain}}.

#INSTRUCTIONS#
1. Analyze {{org_domain}} to understand the company's products and target market
2. Search for companies offering similar solutions in the same industry
3. Focus on direct competitors, not just companies in the same broad category
4. Provide reasoning for why each competitor was selected
5. Include competitor websites if available

#OUTPUT FORMAT#
Return competitor names, websites, and brief reasoning for each selection.

Example 3: Security Incident Research

#CONTEXT#
You are tasked with finding recent security or compliance incidents that may indicate a company's IT infrastructure is vulnerable or in need of modernization.

#OBJECTIVE#
Search for mentions of data breaches, compliance fines, IT security incidents, or outdated systems at the company with domain {{org_domain}}.
If none are found, provide a recent example from a similar company in the same industry to highlight common risks.

#INSTRUCTIONS#
1. Prioritize sources from the last 12-18 months
2. Acceptable sources: news articles, compliance reports, industry blogs, press releases
3. Capture: URL, Title, Summary (2-3 sentences)
4. If no specific incidents found for this company, look for relevant industry examples

#OUTPUT FORMAT#
URL: [link]
Title: [title]
Summary: [summary]

Example 4: Funding and Partnership Research

#CONTEXT#
You need to determine if a company has recently raised funding or announced a partnership.

#OBJECTIVE#
Determine if the company with domain {{org_domain}} has had a recent funding round or partnership announcement. Provide verifiable details.

#INSTRUCTIONS#
1. Use the provided {{org_domain}} as the search criterion
2. Perform a web search to find news articles, press releases, or blog posts that specifically mention recent funding or partnerships
3. Only consider information from reliable, verifiable sources
4. If found, record URL, summary, and title
5. Return "No funding or partnership announcement found" if no suitable information is located

#REMINDERS#
- Ensure information is factual and verifiable
- If multiple announcements exist, choose the most recent

Success Tips

1. Start Simple: Begin with basic research tasks and gradually add complexity
2. Test Your Variables: Make sure you're using the correct variable names
3. Be Specific: Vague prompts lead to vague results
4. Include Examples: Show the AI exactly what format you want
5. Set Boundaries: Clearly state what sources to use or avoid
6. Handle Edge Cases: Specify what to return if no information is found

Common Mistakes to Avoid

• Using undefined variables (check the available variable list)
• Being too vague about data sources
• Not specifying output format clearly
• Asking for information that requires real-time access
• Not handling cases where information might not be available
• Mixing multiple unrelated research objectives in one prompt

Advanced Techniques

1. Using Conditional Logic

If specific security incidents are found, provide detailed information;
If none are found, search for similar cases in the same industry as reference.

2. Setting Confidence Scores

Provide a confidence score from 1-100 for your classification, where:
- 90-100: Very certain, with clear evidence
- 70-89: Fairly certain, with sufficient evidence
- 50-69: Moderately certain, with some evidence
- 30-49: Not very certain, with limited evidence
- 1-29: Very uncertain, with almost no evidence

3. Multi-step Verification

1. First search the company website to confirm basic information
2. Then search third-party news sources for verification
3. Finally cross-reference multiple sources to ensure accuracy

4. Handling Data Quality

If information sources are unreliable or outdated, note in the results:
- Reliability rating of the information source
- Timeliness of the information
- Whether further verification is needed

Getting Help

If your prompts aren't working as expected:

1. Check Variable Format: Make sure all variables use the {{}} format
2. Verify Output Settings: Ensure your output fields match what you're asking for in the prompt
3. Confirm Information Availability: Verify that the information you're requesting can be found through web search
4. Simplify and Test: Start with simpler prompts first, then add complexity
5. Check Instruction Clarity: Make sure your instructions are clear and step-by-step

Common Use Case Templates

Basic Company Information Research

#CONTEXT#
You are a research analyst who needs to gather basic company information.

#OBJECTIVE#
Collect basic business information about {{org_company_name}} ({{org_domain}}).

#INSTRUCTIONS#
1. Visit the company website to understand their business scope
2. Determine their main products or services
3. Identify their target customer base
4. Find company size and location information

Market Positioning Analysis

#CONTEXT#
You are analyzing a company's market positioning and competitive environment.

#OBJECTIVE#
Analyze {{org_company_name}}'s positioning in their industry.

#INSTRUCTIONS#
1. Identify the specific industry segment the company operates in
2. Find their main competitors
3. Analyze the company's unique value proposition
4. Assess their market position (leader/challenger/follower)

Technology Stack Research

#CONTEXT#
You need to understand the technologies and tools a company uses.

#OBJECTIVE#
Identify the main technology stack used by {{org_company_name}}.

#INSTRUCTIONS#
1. Analyze the company website's technical implementation
2. Search for technical blogs or engineering team information
3. Look for job postings that mention technical requirements
4. Identify main programming languages, frameworks, and tools

For Complete Beginners

If you've never written any kind of prompt before, here's a step-by-step approach:

Step 1: Start with a Simple Question

Think of one specific thing you want to know about a company. For example:

• "What does this company do?"
• "Who are their competitors?"
• "Have they raised money recently?"

Step 2: Write It Like You're Talking to a Person

Find out what {{org_company_name}} does by looking at their website {{org_domain}}.

Step 3: Add More Details

#CONTEXT#
You are helping me research companies.

#OBJECTIVE#
Find out what {{org_company_name}} does by looking at their website {{org_domain}}.

#INSTRUCTIONS#
1. Go to their website
2. Read about their products or services
3. Tell me what industry they're in
4. Explain what they do in simple terms

Step 4: Set Up Your Output Settings

• Field Name: company_description
• Type: Text
• Description: A simple explanation of what the company does

Step 5: Test and Improve

Run your prompt and see what you get. If it's not quite right, adjust your instructions to be more specific.

Remember

• Good prompts lead to good results - Take time to craft clear, specific instructions
• Start simple and build up - Don't try to do everything in your first prompt
• Test with real examples - Use actual company information to test your prompts
• Be patient - It may take a few tries to get exactly what you want

The AI Research tool is powerful, but it needs clear instructions to give you the best results. Think of it as training a very capable research assistant - the clearer your instructions, the better the results you'll get.