Writing Effective Prompts

Master the art of writing effective research prompts in Raycaster AI Workspace

The Art of Prompt Writing

Writing effective prompts is crucial for getting high-quality, relevant research results from Raycaster’s AI. A well-crafted prompt acts as a clear set of instructions that guides the AI to find and analyze exactly the information you need.

Prompt Structure Basics

An effective research prompt typically includes these elements:

1. Context

   Provide relevant background information about what you’re researching

   • Reference the subject (company, person, etc.)
   • Include what you already know about the subject
   • Example: “For , a biotech company focused on gene therapy…”
2. Specific Request

   Clearly state what information you’re looking for

   • Be explicit about the data points needed
   • Use direct questions or imperatives
   • Example: “Identify their main therapeutic areas and most advanced pipeline candidates.”
3. Output Format

   Indicate how you want the information presented

   • Specify structure (paragraphs, bullet points, etc.)
   • Note any categorization or organization preferences
   • Example: “Organize findings by therapeutic area, with bullet points for each pipeline candidate.”
4. Constraints and Scope

   Set boundaries for the research

   • Specify time frame (recent developments, historical data, etc.)
   • Indicate information priority
   • Example: “Focus on developments from the past 2 years, prioritizing clinical-stage programs.”

Example Well-Structured Prompt

Research {company_name} based on their website {company_url} and public information. 
Provide a comprehensive analysis of their cell therapy pipeline, focusing on:
1. Specific cell types being developed (CAR-T, NK cells, etc.)
2. Target indications and current development stages
3. Key distinguishing technologies or approaches
4. Recent clinical trial results (past 18 months)

Format the response as bullet points organized by program, with the most advanced 
programs listed first. Include specific data points like target antigens and trial 
identifiers when available.

Common Prompt Patterns

These templates can be adapted for different research needs:

1. Comparative Analysis

Compare {company_name} with its top 3 competitors in the {specific_field} space. 
For each company, analyze:
- Core technology platforms
- Key differentiators
- Market focus and target customers
- Development stage and commercial readiness

Present the information in a structured format that highlights similarities and 
differences, with a focus on competitive advantages or disadvantages.

2. Detailed Profile

Create a comprehensive profile of {person_name} based on their LinkedIn profile 
{linkedin_url} and public information. Include:
- Career trajectory and key roles
- Areas of expertise and specialization
- Educational background and notable credentials
- Research interests and contributions
- Speaking engagements and thought leadership

Organize the information chronologically and highlight experience specific to {industry}.

3. Technology Assessment

Evaluate {company_name}'s technology platform in {specific_field}. Research:
- Core technological approach and scientific basis
- Key innovations or proprietary methods
- Advantages over traditional/alternative approaches
- Technical limitations or challenges
- Supporting evidence (clinical/preclinical data, publications)

Format as an analytical assessment with evidence-based conclusions about the 
technology's potential impact and viability.

4. Market Positioning

Analyze {company_name}'s market positioning in the {specific_market} landscape.
Determine:
- Target market segments and customer base
- Value proposition and messaging
- Pricing strategy (if available)
- Distribution channels and go-to-market approach
- Key partnerships or collaborations

Present findings as a structured market analysis with emphasis on how they 
differentiate from competitors in the same space.

Writing Prompts for Different Column Types

Adapt your prompt style to match the column type for optimal results:

TEXT Column Prompts

For text columns, be specific about the format and depth:

Short Text Example

Provide a one-sentence description of {company_name}'s core business focus 
based on their website {company_url}.

Long Text Example

Write a detailed analysis (3-4 paragraphs) of {company_name}'s approach to 
drug discovery based on their website, publications, and public information. 
Cover their technology platform, target selection strategy, and key differentiators.

NUMBER Column Prompts

For number columns, specify exactly what figure you need and its format:

Determine {company_name}'s total funding amount in US dollars based on public 
funding announcements, investor information, and financial news. Provide only 
the numeric value without currency symbols or formatting.

BOOLEAN Column Prompts

For boolean columns, frame a clear yes/no question:

Determine whether {company_name} has any active clinical trials for oncology 
indications. Base your assessment on their pipeline information, clinicaltrials.gov, 
and recent announcements. Answer only yes or no.

SINGLE_SELECT Column Prompts

For single-select columns, emphasize the need to choose exactly one option:

Based on {company_name}'s product portfolio and pipeline, determine their 
PRIMARY therapeutic focus area. Select EXACTLY ONE option from the available 
choices, choosing the area that represents their main commercial or development focus.

MULTI_SELECT Column Prompts

For multi-select columns, clarify that multiple selections are possible:

Identify ALL therapeutic areas where {company_name} has active programs or 
products according to their website {company_url} and pipeline information. 
Select ANY applicable areas from the available options. Select NONE if no 
listed areas apply.

Using Variables Effectively

Variables make your prompts dynamic by referencing data from other columns:

1. Insert Variables

   Use curly braces to include variables:

2. Combine Multiple Variables

   Reference several data points for comprehensive research:

   Research {company_name}'s activities in {therapeutic_area} based on their 
   website {company_url} and recent publications by their scientists.
   

3. Handle Missing Values

   Account for potentially missing data with conditional instructions:

   Analyze {company_name}'s approach to {research_area}. If {research_area} 
   is not specified, research their main technology platforms instead.
   

Commonly Used Variables

In Company Sheets

• {company_name} - Company’s official name
• {company_url} - Company’s website URL
• {industry} - Company’s industry classification
• {founding_year} - Year the company was founded
• {headquarters} - Company headquarters location

In People Sheets

• {person_name} - Individual’s full name
• {linkedin_url} - LinkedIn profile URL
• {title} - Current job title
• {company} - Current company affiliation
• {education} - Educational background

Special Variables

• {search_keywords} - Keywords from your organization settings
• {organization_name} - Your organization’s name
• {current_date} - Today’s date

Prompt Optimization Techniques

Be Specific, Not Vague

❌ Vague

”Tell me about ‘s products.”

✓ Specific

”Identify ‘s three most advanced product candidates, including their development stage, target indication, and mechanism of action.”

Use Structured Requests

❌ Unstructured

”Research ‘s technology and markets and partnerships and future plans.”

✓ Structured

“Research and provide:
1. Core technology platforms
2. Target market segments
3. Key strategic partnerships
4. Announced future development plans”

Clarify Output Format

❌ Ambiguous Format

”Analyze ‘s competitive position.”

✓ Clear Format

”Analyze ‘s competitive position using a SWOT format (Strengths, Weaknesses, Opportunities, Threats), with 2-3 bullet points for each category.”

Troubleshooting Prompts

If you’re not getting the results you expect, try these adjustments:

1. Add More Specificity

   Include additional details or parameters to focus the research

2. Break Down Complex Requests

   Split one complex prompt into multiple simpler ones

3. Adjust Scope

   Narrow or broaden the focus area based on available information

4. Check Variables

   Ensure all variables in your prompt have corresponding data

5. Enable Web Search

   For recent or specialized information, make sure web search is enabled

Prompt Debugging Examples

Original Prompt (Too Broad):

“Research ‘s technology platform.”

Improved Prompt:

“Research ‘s gene editing technology platform, specifically focusing on their CRISPR-based approaches. Analyze: (1) their specific technical innovations, (2) key differentiators from other CRISPR companies, (3) intellectual property position, and (4) most advanced applications currently in development.”

Original Prompt (Ambiguous):

“Is successful?”

Improved Prompt:

“Evaluate ‘s market success based on objective metrics: (1) funding raised to date, (2) progress of lead programs through clinical development, (3) significant partnerships or collaborations, and (4) comparative position against direct competitors. Provide a fact-based assessment with supporting evidence for each metric.”

Remember that prompt writing is an iterative process. Test different approaches and refine your prompts based on the results you receive. For specific research tools and column types, see Research Types.

Assistant

Responses are generated using AI and may contain mistakes.