[vc_row][vc_column][vc_column_text css=””]
📝 Developing an AI Co-Pilot for LinkedIn Automation
What is the AI Co-Pilot?
The AI Co-Pilot is a 10-part automated AI system that takes a complete clone of your voice, fed in by your own LinkedIn history and data. It analyzes various aspects of your tone of voice, including:
- Vocab and word choice
- Grammatical patterns
- Punctuation
- Sentence structure and length
- Rhetorical devices
- Paragraph structure
- Mood, coherence, and cohesion
- Idiosyncrasies and quirks
- Figurative languages
How Does it Work?
The AI Co-Pilot works by:
- Feeding in your comment history on LinkedIn
- Creating a document that breaks down your tone of voice into 10-11 different components
- Using this document to generate comments that sound like you
Key Components
- Turn of Voice Document: a document that breaks down your tone of voice into 10-11 different components
- Comment Generator: a system that generates comments based on your turn of voice document and LinkedIn post information
- Web Hook: a custom hook that sends information from LinkedIn to the AI Co-Pilot
Setting Up the AI Co-Pilot
To set up the AI Co-Pilot, you need to:
- Create a web hook
- Feed in your comment history on LinkedIn
- Create a turn of voice document
- Configure the comment generator to your personal preferences
Benefits
The AI Co-Pilot can help you:
- Save time by automating comments on LinkedIn
- Increase engagement and conversions on your posts
- Build your personal brand and reputation on LinkedIn
Example Output
Here’s an example of a comment generated by the AI Co-Pilot:
“How quick is that? I love your approach to addressing demand curves and profit margins, it’s all about balance and urgency and value. Have you found any particular strategies that consistently increase conversion rates?”
Code Snippet: Setting up the Web Hook
// Create a custom web hook
// Choose a hook and give it a name (e.g. "LinkedIn Takeover")
// Click "Save" to generate the web hookNote: This is a simplified example and actual code snippet may vary depending on the implementation.## Webhook Setup 🌐
To set up the webhook, follow these steps:
Changing the Webhook URL
- Right-click on
content script.jsand select “Open with” → “Text Edit” (on Mac) or the specified text editor (on Windows) - Replace the existing webhook URL with the new one:
eu2 - Save the file and close it
Loading the Extension 🔌
- Go to Chrome extensions by clicking on the three vertical dots in the top right corner and selecting “More tools” → “Extensions”
- Enable developer mode by toggling the switch in the top right corner
- Click “Load unpacked” and select the folder containing the extension
- Make sure the extension is pinned in the toolbar for easy access
Validating the Webhook 📝
- Run the scenario again and wait for new data to arrive
- Refresh the LinkedIn page and click on the “Add comment” section
- The post text and comments should be fetched successfully, indicating that the webhook is working correctly
Creating a Custom Tone of Voice 📝
To create a custom tone of voice, follow these steps:
Exporting LinkedIn Data 💻
- Go to LinkedIn settings and click on “Data privacy” on the left-hand side
- Click on “Get a copy of your data” to export all your post data
- Request a download, which usually takes around 10 minutes
Analyzing the Data with GPT Modules 🤖
- Create a clone of your tone of voice using the GPT modules
- Analyze the text to create tone of voice guidelines for vocab and word choice
- Provide detailed guidelines for each element, including relevant examples from the given text
GPT Module Example 📝
Analyze the given text for the following:
- Unique or repetitive words and phrases
- Frequency and diversity of vocabulary
- Use of jargon, slang, or colloquialisms
- Distinctive or unusual word choices
- Reading age level required to understand the vocabulary used
Accessing the Blueprint 📁
- Go to the school community and access the classroom
- Search for “turn of voice” to find the segmentation prompt and blueprint
- Use this blueprint to create your custom tone of voice
Creating an Airtable 📊
- Create a new airtable to store your data
- Start from scratch and design the table according to your needs
Here is an example of what the airtable might look like:
| Column Name | Description |
|---|---|
| Vocab and Word Choice | Guidelines for vocab and word choice |
| Grammatical Patterns | Guidelines for grammatical patterns |
| Tone and Style | Guidelines for tone and style |
| … | … |
Note: This is just a sample table, and you can customize it according to your needs.## Build Together LinkedIn Voice with Airtable and OpenAI 🚀
Creating a Custom Turn-of-Voice Agent
To create a custom turn-of-voice agent, you’ll need to follow these steps:
- Download data from various sources, including LinkedIn data
- Create a new base in Airtable with a beautiful name, such as “Build Together LinkedIn Voice” 💬
- Add fields to the base, including:
- Who (with a person Emoji: 👨)
- Turn off voice (with a blue heart Emoji: 💙)
- Created time (with a clock Emoji: ⏰)
- Last modified time (with a clock Emoji: ⏰)
Setting up Airtable
- Right-click on the “Edit Field” button and select “User” to add the created time and last modified time fields
- Use Emojis to make the data more visually appealing
- Get 10,000 free records with Airtable
Integrating with OpenAI
- Create a new completion in OpenAI with a max token limit of 0 (default to maximum)
- Add a message with the role as “user” and the message content as the prompt
- Use the same prompt all the way through, except for the aspect of grammar and tone of voice that changes in each module
Why 10 Individual Modules?
- Quality output will be better with 10 individual modules
- Each module has a specific mission, such as looking at vocab and word choice
- Renaming each module to reflect its specific mission
Creating Multiple Modules
- Copy and paste the information to create multiple modules
- Rename each module accordingly
Framework for Analysis
- VUP (Vocab and Understanding Patterns)
- Grammatical Patterns: analyze the writer’s use of specific grammatical structures, verb tenses, and consistency
- Evaluate the writer’s ideas and how they flow and connect throughout the piece
Importing and ExportingBlueprints
- Download the blueprint as a JSON file
- Import the blueprint into the scenario
- Select the file and click “Save” to populate the entire thing automatically
Conclusion
- Repeat the process for all 10 modules
- Save time by importing the blueprint and letting it populate automatically## 🤖 LinkedIn Automations with AI Co-pilot 💻
Evai: LinkedIn AI Business
- Evai is a LinkedIn AI business founded by Jo Alab, a LinkedIn expert who has developed cool LinkedIn automations.
- Check out Evai for interesting features to upgrade your LinkedIn game.
Focus on One Platform
- To achieve success, focus on one platform at a time, whether it’s YouTube, LinkedIn, or another platform.
Google Docs for Automation 📝
- Create a Google Doc to organize information and feed it into a language model to create an incredible automation.
- Document structure:
- Vocab and Word Choice
- Grammatical Patterns
Turn of Voice 10p Part System 🗣️
- A system that allows AI-generated content to sound like an individual, not a robot.
- Uses idiosyncrasies, structures, and coherence to create a natural-sounding tone.
London Meetup and AI Event 🇬🇧
- Attend AI events to network and learn from experts in the field.
Google Docs Creation
- Create a new document and name it (e.g., “Turn of Voice”).
- Add content to the document, including vocab and word choice, grammatical patterns, and more.
Airtable Upload 📊
- Upload the turn of voice to Airtable.
- Use a random turn of voice as an example if needed.
Scenario Running ⏃♂️
- Run the scenario to check if everything is working correctly.
- Airtable will grab the vocab and word choice, creating an interesting piece of text.
Turn of Voice Analysis 🤔
- Analyze the result of the scenario run.
- Check the active voice, punctuation, and usage of ellipses.
Document Formatting 📄
- Tidy up the document by adding formatting steps and tightening up the outputs.
Smart Scenario Creation 📊
- Create a new smart scenario to continue building the AI co-pilot.## Automating LinkedIn Comment Responses with ChatGPT 🔥
Understanding the Scenario
In this scenario, we’re using ChatGPT to generate responses to LinkedIn comments based on a provided tone of voice and a dataset of comments and posts.
Setup
We’ll be using ChatGPT 4, which has shown impressive results in terms of language reliability and memory.
User Message Content
We’ll be using the following prompt:
You are an AI agent
- You have three LinkedIn comment responses in the attached tone of voice
- Your primary goal is to create engaging, authentic comments that align with the provider’s tone of voice and contribute meaningfully to the conversation
Important Instructions
Read and internalize the separate tone of voice document provided Ensure all generated content adheres to this specified tone
Response Guidelines
- Generate exactly three responses for the post
- One short response (3-8 words)
- Two mid-length responses (2-4 sentences)
- Add value to the post by:
- Providing a unique perspective
- Asking questions
- Referencing specific points in the post
- Including follow-up questions
- Ensuring responses are diverse and contextually relevant
Using ChatGPT
We’ll paste the prompt into the automation tool and set up the response format to JSON object.
Advanced Settings
| Setting | Value |
|---|---|
| Tokens | 0 |
| Response Format | JSON object |
| Pass JSON Response | Yes |
Running the Scenario
Once set up, we’ll run the scenario and get our responses. We can then copy and paste the responses into LinkedIn.
Example Responses
- “Love that! Consistency is key 💯”
- “What an inspiring story! Her perseverance and dedication are truly admirable 😊”
- “This really resonates with me. International students face immense challenges, but being persistent and dedicated to their roles… what steps can we take to support them?”
[/vc_column_text][/vc_column][/vc_row]
