How To Automate Your Tweet Generation Workflow Using LLMs

Discover how local LLMs can revolutionize your tweet generation process and boost your productivity!

1 min read
productivity genai

All the tools buffer, tweethunter, hyperfury charge money to generate tweets.

I had an itch to try running local llms and play with it.

I had also been re-reading Content OS by Justin Welsh. He talks about generating content with various perspective about the same content. So I thought, I can automate a bunch of it by using prompt templates, inspiration tweet examples and mistral 7B.

Packages & Tooling:

  • Vanilla Python

  • notion-sdk-py

  • LlamaIndex

  • Ollama

  • Mistral 7B

  • #### Logic workflow CleanShot 2024-02-09 at 16.14.01.png

Code is at https://github.com/nehiljain/tweet-automation

Quick Introspection:

  • Coding difficulty: Low
  • Time to develop end to end: 6 hr for V1
  • Usefulness: Instantly I was able to schedule 5 tweets in 15 mins
  • Enjoyment: Felt great playing with Notion, Ollama, Obsidian

[!Ship30for30 advice] But once you start writing, and building your library, you start to build momentum.

  • You write...
  • Which creates data points...
  • Which reveals patterns (what works/what doesn’t)...
  • Which shows you where there are opportunities to double-down...
  • Which makes it easy to write the next thing...
  • And the next thing...

And so on...

This is the game of Digital Writing.