> ## Documentation Index > Fetch the complete documentation index at: https://docs.raycaster.ai/llms.txt > Use this file to discover all available pages before exploring further. # Setting Up Research Columns > Configure columns to automatically gather intelligence with AI

Research Column Configuration

Set up intelligent columns that automatically gather and analyze the exact information you need.

Customizable Research

Define exactly what information to gather with custom prompts and parameters

Format Control

Choose how research results appear in your columns, from text to numbers to selections

Automated Workflows

Configure once, then gather intelligence automatically across your entire dataset

## Configuration at a Glance
Configuration Step
Purpose
1. Enable AI Capabilities
Turn on AI research for a column to unlock intelligent data gathering
2. Select Research Tool
Choose the specialized research tool that matches your data needs
3. Write Research Prompt
Provide clear instructions for what information to gather and analyze
4. Select Input Variables
Choose which columns provide context for the research operation
5. Configure Settings
Fine-tune web search, formatting, and other research parameters

Research columns operate in the context of your sheet's data. Make sure your sheet has the necessary input columns (company name, URLs, etc.) that research operations will reference.

## Configuring Columns for AI Research One of Raycaster's most powerful features is the ability to configure columns to automatically gather and analyze information through AI research. This guide explains how to set up and optimize research-enabled columns. ## Enabling Research for a Column

Navigate to the sheet where you want to add research capabilities, then:

Column Menu
Manage Columns

Or click the "+" button to add a new column, then configure it.

In the column editor, find the "AI Capabilities" toggle:

Enabled

This unlocks the research configuration options below.

Choose the type of research tool that best matches your needs:

STRUCTURED
COMPANY\_ENRICHMENT
PEOPLE\_ENRICHMENT
PATENT\_RESEARCH

Each tool is optimized for different types of data. STRUCTURED works well for most general cases.

Provide clear instructions for what information to gather:

Write a concise summary of {company_name}'s business model based on their website {company_url} and recent news. Focus on their main products or services, target market, and revenue streams.

Choose which columns provide context for your research:

company\_name
company\_url
Web Search Enabled

Enabling web search allows the AI to gather current information from the internet.

Video tutorial coming soon

 ## Selecting the Right Research Tool Raycaster offers specialized research tools for different needs: | Research Tool | Purpose | Best For | Compatible Column Types | | ----------------------- | ---------------------------------------------------- | ------------------------------ | ------------------------------------- | | **STRUCTURED** | General-purpose research with formatted output | Most research needs | TEXT, NUMBER, BOOLEAN, SELECT columns | | **ADVANCED** | In-depth analysis with comprehensive data collection | Detailed research requirements | Most column types | | **EMAIL** | Email address discovery and verification | Finding contact information | TEXT, EMAIL columns | | **COMPANY\_ENRICHMENT** | Comprehensive company data gathering | Complete company profiles | ACTION columns | | **PEOPLE\_ENRICHMENT** | Detailed professional background analysis | Complete people profiles | ACTION columns | | **LINKEDIN\_POSTS** | Extract insights from LinkedIn content | Social media analysis | ACTION columns | | **FETCH\_PEOPLE** | Discover team members and contacts | Building contact lists | ACTION columns | Choose the tool that most closely matches your specific research objective. For general needs, STRUCTURED research works well across most situations. ## Writing Effective Research Prompts

Common Prompt Mistakes

  • • Being too vague ("research this company")
  • • Forgetting to specify format ("list key information")
  • • Not referencing variables properly ({'{variable_name}'})
  • • Requesting too many things in one prompt

Prompt Best Practices

  • • Be specific about exactly what information you need
  • • Indicate your preferred format and structure
  • • Reference existing columns with {'{column_name}'}
  • • Focus each prompt on a single research objective
### Prompt Structure Formula

\[Action verb] + \[What information] + \[For what entity] + \[Using what sources] + \[In what format]

Example:

Analyze the funding history for {'{company_name}'} based on their website {'{company_url}'} and public funding databases. Provide a chronological list of funding rounds with date, amount, and lead investors.

### Prompt Examples by Column Type
GOOD EXAMPLE
Write a concise overview of {'{company_name}'}'s business model based on their website {'{company_url}'} and recent news. Focus on their main products or services, target market, and revenue streams. Structure as 2-3 paragraphs with bullet points for key offerings.
Why it works: Specifies exactly what information to gather, references input variables properly, and indicates the desired format and structure.
GOOD EXAMPLE
Determine the most recent employee count for {'{company_name}'} based on LinkedIn data, their website {'{company_url}'}, and other public sources. Return only the numeric value representing the total number of employees. If a range is provided, use the midpoint.
Why it works: Clearly specifies a numeric output, provides guidance on how to handle ranges, and identifies the specific data point needed.
GOOD EXAMPLE
Determine whether {'{company_name}'} has received venture capital funding in the last 3 years based on their website {'{company_url}'}, press releases, and funding databases. Return TRUE if they have received VC funding since \[current\_year-3], otherwise return FALSE.
Why it works: Asks a clear yes/no question with specific criteria for determining the answer, and explicitly requests a boolean response.
GOOD EXAMPLE
Based on {'{company_name}'}'s website {'{company_url}'} and public information, determine their primary therapeutic focus area. Choose only ONE option from: Oncology, Neurology, Immunology, Cardiology, Rare Disease, or Other. If multiple areas apply, select the one with the strongest emphasis.
Why it works: Lists all available options, emphasizes that only one selection is needed, and provides guidance for handling edge cases.
## Selecting Input Variables Input variables provide context for your research prompt: In the column configuration, find the "Input Variables" section Choose which columns should be referenced in your prompt: * Company sheet variables: company\_name, company\_url, industry, etc. * People sheet variables: person\_name, linkedin\_url, title, etc. * Special variables: search\_keywords, organization\_settings, etc. Make sure all variables referenced in your prompt are selected here * The system will warn you if you're missing any variables * Variables used in the prompt are automatically highlighted ### Common Input Variables #### Company Sheet Variables * **company\_name**: The name of the company * **company\_url**: The company's website URL * **company\_domain**: Just the domain portion of the URL * **industry**: The company's industry classification * **location**: Company headquarters location * **company\_linkedin\_url**: The company's LinkedIn profile #### People Sheet Variables * **person\_name**: The individual's full name * **person\_linkedin\_url**: Their LinkedIn profile URL * **title**: Their job title * **company**: Their company affiliation * **email**: Their email address (if available) #### Special Variables * **search\_keywords**: Keywords defined in your organization settings * **organization\_info**: Information about your organization * **current\_date**: The current date for time-sensitive research ## Configuring Web Search Enable or disable web search based on your research needs: In the column configuration, find the "Web Search" toggle Toggle web search ON or OFF: * ON: The AI will search the internet for information (recommended for most cases) * OFF: The AI will use only the provided variables and internal knowledge ### When to Enable Web Search * **Enable when**: * You need the most current information * Your research involves recent events or news * You're researching specific details not likely in the AI's knowledge base * You need verification from multiple sources * **Disable when**: * You want faster processing * You only need analysis of already-provided data * You're working with confidential information * You need consistent results regardless of when research is performed Web search increases processing time but generally provides more comprehensive and current results. For time-sensitive information like funding, news, or recent developments, web search is strongly recommended. ## Advanced Configuration Options ### Output Formatting For TEXT columns, you can specify formatting preferences: 1. **Structure**: Choose from paragraph, bullet points, or numbered list 2. **Length**: Set approximate target length (short, medium, long) 3. **Style**: Select analytical, descriptive, or concise 4. **Highlight Important Points**: Toggle to emphasize key findings ### Processing Priority For batch research operations, set column priorities: 1. Higher priority columns are processed first 2. Set priority from 1-5 with 1 being highest priority 3. This affects the order of processing, not the quality of research ### Source Requirements Specify source requirements for verification: 1. **Minimum Sources**: Set the minimum number of sources required (1-5) 2. **Source Types**: Specify allowed source types (official websites, news, academic, etc.) 3. **Source Recency**: Set maximum age for sources (any, last year, last 6 months, etc.)

Once you've configured a research column, you can start researching by clicking the research button in individual cells or by using batch research for multiple rows. For more details, see Viewing Research Results.