Understanding Prompts
In Kakiyo, prompts are instructions that guide how the AI communicates with your prospects on LinkedIn. Think of prompts as the "script" or "playbook" that teaches the AI how to have effective conversations on your behalf.
The Structure of Prompts
Each prompt in Kakiyo consists of two primary sections:
1. Context
The Context section provides overall guidance for the entire conversation. This includes:
The AI's role in the conversation
Campaign objectives (booking meetings, sharing content, etc.)
Conversation style and tone preferences
Word count limitations
General instructions for handling prospect responses
Special cases handling (objections, frequently asked questions, etc.)
2. First Message
The First Message section specifically instructs the AI on how to craft the initial outreach message. This is critical because:
First impressions determine whether prospects will engage
Initial messages require a different approach than follow-up conversations
This message needs to be particularly compelling to stand out
Prompt Variables
Prompts can contain variables, which are placeholders that get replaced with specific content when the campaign runs:
Required Variables
Two variables are mandatory for all prompts:
{{productDescription}}- Information about your product or service{{prospectDescription}}- Details about the prospect being contacted
Important: You don't need to fill in these required variables manually. The system automatically populates them with the appropriate information during campaign execution.
Optional Variables
The system provides several optional variables you can use in your prompts:
{{agentName}}- The name of the LinkedIn profile owner{{prospectName}}- The name of the prospect being contacted{{companyName}}- The company name of the prospect
Important: These variables are also filled automatically by the system. You can choose to use them or not in your prompt - they're entirely optional.
Custom Variables
You can create additional custom variables to make your prompts more flexible:
{{goal}}- Different campaign objectives{{goalLink}}- Links to specific landing pages{{fallback}}- Alternative offers for uninterested prospects{{language}}- Language preferences for international campaigns
Important: Unlike required and optional variables, custom variables must be created by you and filled in manually when setting up your campaign. This gives you the flexibility to reuse prompt templates across different campaigns by changing only the specific values.
AI Models
When creating a prompt, you'll select which AI model powers your conversations. This choice affects both performance and credit consumption:
Recommended Models
GPT-4.1: Good balance between performance and credit usage
Claude 3.7 Sonnet: Strong conversational abilities with reasonable credit consumption
GPT-4o: Reliable performance with moderate credit usage
Other Available Models
Claude 3.5 Haiku
OpenAI o1
OpenAI o3-mini
GPT-4o mini
GPT-4.5
GPT-4.1 mini
GPT-4.1 nano
⚠️ Warning about GPT-4.5: This model consumes approximately 100-150 times more credits than other models. Use with caution as credits can be depleted extremely quickly.
Important Credit Considerations
Different models consume different amounts of credits per action
For most LinkedIn prospecting conversations, the recommended models provide excellent results without excessive credit consumption
Consider testing different models on similar prompts to find the best balance of performance and credit efficiency for your use case
Every AI action in Kakiyo (generating messages, testing prompts, etc.) consumes credits from your account based on the AI model used. For more information, see Understanding Credits in Kakiyo.
Benefits of Well-Crafted Prompts
Consistency: Ensure all prospect conversations follow your preferred approach
Personalization: Create natural-sounding messages tailored to each prospect
Optimization: Refine your messaging based on what generates the best responses
Scalability: Apply successful conversation strategies across multiple campaigns
Prompt Best Practices
Be specific about the conversation goal
Include clear instructions about tone and style
Set word count limitations for messages
Specify how to handle common objections
Provide guidance on when to suggest meetings
Include instructions for keeping conversations natural
Next Steps
Once you understand what prompts are, you can:
Learn how to create and test prompts
Explore prompt variables in depth