Twinkling

Inspiration

I have 1600 notes that are kind of the essence of what I learned over the last year and would love to share it with others! These notes contain valuable insights and knowledge, but they were just sitting there, not being utilized to their full potential. I wanted a way to share these bits of wisdom with the world effortlessly.

What it does

It scans through your notes, selects interesting snippets, and transforms them into engaging tweets. This way, you can share your knowledge and insights with your followers without the constant effort of crafting new content.

How we built it

We developed Twinkling using Python, leveraging several key components:

1. A Note Manager to access and process your existing notes
2. A Model Manager utilizing AI to generate tweet-worthy content
3. A Content Generator that transforms your notes into concise, engaging tweets
4. A Twitter integration module for seamless posting We also implemented a user-friendly interface that allows you to review and approve tweets before they're posted.

Challenges we ran into

using Twitter API, most of the content on the internet is about V1.1 while now there is V2, also Twitter added a lot of restriction about accessing tweets.

Accomplishments that we're proud of

Building a useful prototype that I'm actually gonna use in a span of 4 hours.

What we learned

We learned how to use Tweepy and set up OAuth for Twitter. We also got better at fixing errors when working with APIs.

What's next for Twinkling

1. Migrate to AWS with automated message at a given time
2. collect stats about tags, keywords, the best time to post and such.
3. Develop a memory system (similar to memGPT) to remember user preferences and improve tweet generation over time
4. Expand to other social media platforms beyond Twitter
5. build RAG pipeline with watchman to auto-update when new notes are added, so I can generate a post about a given topic

Built With

• llama-3.1
• python
• sambanova
• tweepy