This document explains the AI Web Research node, which combines web search and structured data extraction capabilities into one node. Built on a proprietary AI model, this node enables automated web research, data analysis, and information synthesis from multiple sources.

​
How to Use the Node

  1. Add your research prompt to the node

    • For best results, use the structure: “Given ___, produce/analyze/research/identify ____”
    • Example: “Given a company’s website, research and identify their main product offerings”

  2. Click on “Generate Inputs and Outputs”

    • This will create custom inputs and outputs based on your research needs

  3. Connect the generated inputs

    • Link required data from previous nodes in your workflow

  4. Review the outputs

    • Citations and reasoning are always included
    • Additional outputs are dynamically generated based on your research prompt

  5. Complete your workflow

    • Connect outputs to subsequent nodes as needed

​
Research Types

​
Quick Research

  • Optimized for identifying discrete values from the web
  • Best for: Numbers, simple classifications, locations, URLs
  • Faster processing time
  • Ideal for specific, fact-based queries

​
Deep Analysis

  • Performs comprehensive research across multiple websites
  • Capable of analyzing long PDFs and complex content
  • Best for: In-depth research, people research, document analysis
  • Provides thorough analysis with broader context

​
Node Outputs

​
Standard Outputs

  • Citations: Sources referenced in the research
  • Reasoning: Detailed explanation of how conclusions were reached

​
Dynamic Outputs

  • Generated based on your research prompt
  • Recommended limit: ≤5 high-quality outputs
  • Customizable through “Generate Inputs and Outputs”

​
Research Examples

​
Example Workflows

​
Company Research

Research Prompt: “Given [startup description] and [target venture fund name], research the investment thesis and determine potential fit”

​
Compliance Analysis

Research Prompt: “Given [company name] and [website], analyze compliance certifications and leadership”

​
Financial Research

Research Prompt: “Given [company name], research and extract key financial metrics from latest 10-K”

​
Healthcare Provider Analysis

Research Prompt: “Given [practitioner website], analyze service offerings and patient accessibility”

​
Real Estate Research

Research Prompt: “Given [location] identify and analyze potential investment properties”

  • View Example Workflow
  • Outputs: Property options, median house price, rental yield, real estate agent websites

​
Additional Automation Examples

​
Lead Enrichment Workflows

  • “Given a company’s LinkedIn profile URL, research their tech stack and recent technology investments”
  • “Given a prospect’s website, identify their sustainability initiatives and ESG commitments”
  • “Given a list of target accounts, research their quarterly earnings announcements and major business changes”

​
Competitive Monitoring

  • “Given our competitor’s product page URL, track new feature announcements”
  • “Given a competitor’s name, find new job postings in specific departments”
  • “Given a list of competitor domains, track their partnership announcements and integrations”

​
Content & SEO Automation

  • “Given our blog post URL, research competing articles and identify content gaps”
  • “Given our target keyword, analyze top-ranking pages’ content structure and sources”
  • “Given our product category, monitor industry forums for common user pain points and requests”

​
Best Practices

​
Writing Effective Prompts

  1. Use the recommended format for intuitive results: 
    "Given [context], produce/analyze/research/identify [specific outcome]"
  2. Be specific about required information
  3. Include key aspects to focus on
  4. Specify any source preferences or requirements

​
Optimization Tips

  • Create separate nodes for more than 5 outputs
  • Use Quick Research for simple data points
  • Use Deep Analysis for comprehensive research
  • Chain nodes for complex research tasks

​
Advanced Usage

​
Research Node Chaining

  • Use multiple nodes for complex research tasks
  • First node: Generate initial options
  • Subsequent nodes: Deep dive into specific aspects
  • Connect outputs to create comprehensive workflows

​
Output Customization

  • Outputs are dynamically generated based on research needs
  • Can be modified to focus on specific aspects
  • Regenerate inputs/outputs by editing the prompt and clicking the generate button again

This node represents the most advanced research capability on the Gumloop platform, combining automated web research with precise data extraction to deliver comprehensive, accurate results.

CategorizerSummarizer