Over the weekend, I combined LLMs (Haiku, Google Flash, and GPT3.5) with OpenAI's text embedding small to tackle the hallucination problem. The result? A game-changing solution for a personalized Twitter feed!

Now, I can easily find the top 20 engage worthy tweets tailored to my needs without endless scrolling. And it's costing me just $0.1/100 tweets processed, with room to reduce costs.

Who else wants a custom Twitter experience like this? MevinX is coming soon to help you with doom scroll.

Over the weekend, I combined LLMs (Haiku, Google Flash, and GPT3.5) with OpenAI's text embedding small to tackle the hallucination problem. The result? A game-changing solution for a personalized Twitter feed!

Now, I can easily find the top 20 engaging tweets tailored to my needs without endless scrolling. And it's costing me just $0.1/100 tweets processed, with room to reduce costs.

Who else wants a custom Twitter experience like this? MevinX is coming soon to help you with doom scroll.

Here’s an experiment that I did it with LLMs and Embedding.

When you want to differentiate yourself from the competitors, you will have to do a lot of experiments on these edge techs. Without which, the chances of building a useful product comes down.

The problem that MevinX mainly solves is to avoid doom scrolling in twitter, and give you focused list of tweets that can help you engage quickly and move on to your goals.

buildinpublic #indiehacking

Over the weekend, I tested LLMs and Embeddings to solve the hallucination problem.

As an extension to last week's experiment. The combination of both is mind blowing!

1. Feature extraction using Google Flash
2. Embedding with OpenAI text embedding small

I have been trying to create an echo chamber for twitter, in a way that is not just curates your "For you" tab with all your interactions, but helps you tune it to your needs.

I used LLM as a feature extractors (extracting topics, sentiment, etc.,)

Embedded all the topics, sentiment, tweets, etc., and ran through a complex logic that will create a weighted average score for each of the tweets from my feed.

Voilà! My feed is exactly as what I need. I can find the top 20 tweets that I can engage with, without the need for reading 200 tweets.

It's currently costing me $0.1/100 tweets crawled, but there's a lot of room to reduce this cost.

This is too much! Seems like google has trained this AI with just troll Reddit and Yahoo answers posts. 🤷 https://t.co/q9cHMmVTHD

No. Twitter.

X is easier when you want to open the website, it's not catchy enough.

One good thing about twitter is that, they don’t notify when a bot follows me. If a real user follows, they notify.

Glad X has figured out something about bot. But why not just block and restrict only notifications?

Many of us here forget the fact that we have to market our product.

Yes! even after a failed product hunt launch.

Prioritize marketing as much as you prioritize building the product.

#buildinpublic #mevinx https://t.co/FyA5wa62XY

What's the best topics classifier? I am using LLM, but the outputs are mostly hallucinations.

Trying to classify text with a range of 100-150 topics.

With MevinX, I stopped doom scrolling twitter. I get top 20 tweets that’s worth interacting with based on my interests. And that’s it! Done in 15 minutes.

Looking for private beta testers. Want access? DM me.

buildinpublic https://t.co/S8W4yCuQRe