Implemented my First GPT and experimenting
LinkedIn Post
Building my LinkedIn GPT Assistant in Just 4 Hours
(More ideas, more focus on writing, better learnings)
I analyzed my LinkedIn writing and turned the insights into a GPT assistant.
Here’s how I feed the GPT building on knowledge from my Second Brain 🧠:
‾‾‾‾‾
↳ Data Consolidation: I transformed my insights on writing into a comprehensive PDF. One source: https://lnkd.in/etdVAX5t (and more in the comments)
↳ Expert Techniques: I incorporated distilled wisdom on compelling headlines from Farouk Bacha (see https://lnkd.in/g9CX4ejB) and 12 great headline tips from Lara Acosta.
↳ Content Compilation: I merged all my previous LinkedIn posts into one document using a ChatGPT generated Python script, turning them into a detailed knowledge base for the GPT.
Initial Experiments with the GPT:
↳ Refined 3 posts by feeding my thoughts about the post.
↳ Extracted all book recommendations shared in my posts.
↳ Generated 30 new content ideas based on my historical topics.
↳ Content Analysis: Prompted the GPT to identify recurring themes.
Prompt Examples:
✪ “List all book recommendations I provided so far.”
✪ “Based on my topics so far, what might be the next topics I should write about? Please provide me 30 more content ideas.”
✪ “Based on my headlines, what are topics I’m writing about and how often do I write about them?”
✪ “What are main insights I shared about productivity and tools, especially when it comes to smart learning and AI? Please list at least 20 hints.”
Benefits I noticed:
✪ The GPT’s capability to suggest future topics helps with a consistent content strategy.
✪ Efficiency and Precision: The system saves significant time by understanding my content preferences without repeated briefings.
Next steps:
↳ Keep the GPT Current: Continuously integrate new knowledge and insights
↳ Metadata Optimization: Intend to utilize the LinkedIn API for better organization and retrieval of post metadata.
↳ Enhance Contextual Understanding: Explore ways to expand the GPT’s context window for richer content generation
Building your own GPT is not just about efficiency; it’s about leveraging compounded knowledge. And it’s fun to work with such a great assistent.
👉 Follow me to dive deeper into AI-driven content creation and learn how to build your ‘Second Brain’ for success on LinkedIn!
Content preparation
I externalized my growing knowledge on writing in a GPT and now have an assistent for my daily LinkedIn posts. Thanks to all my data prepared in my Second Brain it was super smooth to feed the briefing data to the GPT.
Steps:
- I did export my insights about writing as a Pdf.
- Based on a distilled version of Magnetic Writing Tips (see comments)
- and my insights from Tired of writing no one wants to read: https://www.linkedin.com/posts/sebastiankamilli_on-writing-activity-7167443208279437313-Bs-d?utm_source=share&utm_medium=member_desktop
- combined with a briefing on writing hook headlines I learned from Farouk Bacha: https://www.linkedin.com/posts/faroukbacha_lara-acosta-linkedin-hook-strategy-activity-7176193307088171008-YgmF?utm_source=share&utm_medium=member_desktop and with 12 tips for writing headlines from Lara Acosta
- I did export all my LinkedIn posts in one big Pdf. GPT provided me with a python script to extract that information from my Second Brain and create that combined Pdf. This became my knowledge base for the GPT
- And I extracted all my post headlines so far and provided that too
First impressions:
- I can now learn more about my writing behavior. Some simple prompts I experimented with:
Based on my headlines what are topics I’m writing about and how often do I write about it?
List all books recommendations I provided so far
What are main insights I shared about productivity and tools. Especially when it comes to smart learning and AI. Please list at least 20 hints.
Based in my topics so far, what might be next topics I should write about? Please provide me 30 more content ideas
- It saves time as I don’t have to provide my instructions again and again
- The initial quality of my refined posts is great to build on
- I’m looking forward to a bigger context window, where I can really build on all my previous posts.
What I want to learn next:
- How to best keep the GPT up-2-date and tune it with new knowledge and my growing posts knowledge base
- How can I increase the context window, so that the GPT can leverage all my insights
- I need to better organize my posts metadata to e.g. easily extract all headlines. I will use the LinkedIn API for that soon
In summary - I can highly recommend playing with building your own GPT. It leads to higher efficiency and leverages compounding knowledge.
👉 Follow me for more insights on working with a Second Brain and start growing your own
Posted on LINKEDIN on 2024-04-24_Wed
Linking
