A good prompt for an AI Phantom follows six parts: objective (what to do), context (background on your campaign), persona (who you are writing as), instructions (which lead fields to use and what to avoid), format (how output should be structured), and tone (voice and style). Start with a clear objective using action words like "generate," "analyze," or "enrich," then add detail in each section. You can also use the built-in prompt templates available in the dropdown menu during setup.
Getting started with AI Phantoms
Before writing your prompt, take a moment to define your goal and pick the right AI Phantom for your task. Once you've selected your Phantom, head to the Prompt section in the setup step and apply the structure below.
Note that you can also use one of the prompt templates suggested when you click the dropdown menu.
How to structure your prompt
Creating the right prompt is key as a well-structured prompt can greatly improve the quality of the AI's output.
Use the structure below to create clear, effective prompts:
1. Objective
Define what the AI should do, using clear action words like: "generate," "enrich," "analyze," or "summarize."
Example:
"Generate five personalized LinkedIn messages for a marketing outreach campaign."
2. Context
Give the AI relevant background so it understands what you're trying to achieve.
Example:
"We are a B2B SaaS company targeting mid-level marketers for a product launch."
3. Persona
Describe who the AI should be writing as, or the tone/angle to take.
Example:
"You are a sales development rep experienced in LinkedIn outreach."
4. Instructions
Show the AI how you'd like the output formatted, or share an example it can follow.
Example:
"Use this format: [Introduction], [Value Proposition], [Call to Action]."
5. Format
Tell the AI how to structure its response.
Example:
"Output the messages in a bulleted list format."
6. Tone
Set the right voice depending on your audience.
Example:
"Use a friendly and professional tone that appeals to mid-level professionals."
Real-world examples from each AI Phantom
Below are full examples using the prompt structure above, tailored to each AI Phantom.
AI LinkedIn Message Writer
Generate personalized outreach messages for your leads.
Prompt example:
- Objective: "Generate five personalized LinkedIn messages for outreach to B2B marketing professionals."
- Context: "We are a B2B SaaS company seeking to engage mid-level marketing professionals to promote our latest lead generation tool."
-
Personas:
My persona: "I'm a LinkedIn outreach specialist targeting B2B SaaS companies."
My audience's persona: "My audience includes mid-level marketing professionals in B2B SaaS companies who focus on customer acquisition and lead nurturing." - Instructions: "Use the recipient's First name, Current job title, Company name, and any Skills to craft personalized messages. Focus on how our tool aligns with their lead generation needs and avoid any pushy sales language."
- Format: "Output a short, concise LinkedIn-style inbox message."
- Tone: "Use a friendly but professional tone, appropriate for mid-level professionals in the SaaS industry."
AI LinkedIn Profile Enricher
Score leads based on their profile quality and engagement potential.
Prompt example:
- Objective: "Analyze LinkedIn profiles to identify and qualify potential leads for our B2B SaaS product."
- Context: "We are focusing on midsized companies with senior marketing professionals, looking for leads that are most likely to engage with our solutions."
-
Personas:
My persona: "I'm a sales development representative qualifying leads for a B2B SaaS solution."
My audience's persona: "My audience consists of senior marketing executives in midsized B2B SaaS companies." - Instructions: "Use the fields First name, Job title, Company name, Skills, and Recent job changes to determine the lead's likelihood to engage. Focus on profiles that have shown recent job changes or growth in the marketing field. Generate a lead qualification score for each profile."
- Format: "Output the analysis as a table with columns for Name, Job title, Company, and Qualification score."
- Tone: "The output should be analytical and neutral, providing a clear indication of engagement potential."
AI LinkedIn Post Responder
Write thoughtful, relevant comments for LinkedIn posts.
Prompt example:
- Objective: "Generate insightful and value-driven comments on recent LinkedIn posts from industry thought leaders in digital marketing."
- Context: "We are engaging with digital marketing thought leaders to build rapport and increase visibility within their networks."
-
Personas:
My persona: "I'm a thought leader in digital marketing, engaging with relevant posts to provide insights and value."
My audience's persona: "My audience includes other thought leaders, their followers, and professionals engaging with their posts." - Instructions: "Use the post content, Author's job title, Company name, and Skills to craft thoughtful and insightful comments. Avoid generic comments like 'Great post!' Instead, reference specifics and share additional insights."
- Format: "Output a short, LinkedIn-style comment no longer than three sentences."
- Tone: "Adopt an insightful and professional tone, with a focus on adding value to the conversation."
Advanced AI Enricher
Add custom enrichment fields to your dataset.
Prompt example:
- Objective: "Enrich a dataset of B2B SaaS professionals by creating custom fields for engagement potential, decision-making authority, and likelihood to convert."
- Context: "We are preparing a lead list for an upcoming product launch targeting senior decision-makers at mid-sized SaaS companies."
-
Personas:
My persona: "I am a sales manager tasked with enriching and qualifying leads for our sales team."
My audience's persona: "My audience includes senior-level marketing and sales professionals who are decision-makers in midsized SaaS companies." - Instructions: "Use the fields Job title, Company name, Current job description, and Skills to create custom fields. Assign a score from 1–10 for Engagement Potential, Decision-Making Authority, and Conversion Likelihood."
- Format: "Output the enriched dataset as a CSV file with columns for each custom score."
- Tone: "The output should be structured and data-driven, suitable for use in a lead scoring system."
Refining your prompt over time
It's okay to start simple and improve over time. If your first prompt doesn't return exactly what you want, tweak one part at a time: change the tone, adjust the format, or provide more context. AI Phantoms improve with clear, consistent input.
Frequently asked questions
Do I have to fill in all six prompt sections?
No. Objective and Instructions are the most important. The other sections (Context, Persona, Format, Tone) improve results but are optional. Start simple and add detail if the output needs refining.
Can I reuse prompts across different AI Phantoms?
The structure works across all AI Phantoms, but the content should be adapted. A prompt for AI LinkedIn Message Writer focuses on message format and personalization, while one for AI LinkedIn Profile Enricher focuses on scoring criteria.
Where do I enter my prompt in the setup?
In the Phantom's setup steps, look for the Prompt section. You can write your own or select from the built-in templates using the dropdown menu.
What if the output is not what I expected?
Tweak one part at a time. Adjust the tone, change the format, or add more context. AI Phantoms improve with clear, specific input. Running a small test batch first helps you iterate without processing your entire list.