Bimoo

Hi @cassie, I've been reading the whitepaper and asked chatGPT to help me succinctly understand it. Two questions:

1. Is this a fair breakdown/summary?
2. It offered the metaphor of "Decentralized AWS" - is this something you're trying to stay clear from? I know you're referencing NoCloud.

/Quilibrium

1874

Victor Mimo

@vic·

Bariloche, Argentina.

Trying this out

Who's built a cool project using LangGraph?

It's so powerful I'm struggling to translate it's architecture into some of my project ideas.

Oh boy, it knows.

Open-source doesn't skip a bit, there's already an alternative to Devin by Cognition AI

https://github.com/stitionai/devika

Gotta say, I'm extremely impressed with Cursor.

I thought there was no way I was paying $20/month for an IDE, but it quickly became a no brainer.

cmd + k is extremely powerful

https://cursor.sh/

Important 🚀

Got around to using SOC last night.
It was shocking how simple the backend seemed.
1. Enter an prompt
2. AI takes screenshot of your screen. Uses GPT4-V to understand
3. Creates followup instructions from there.

It was wonky, but a powerful modality is here.

https://github.com/OthersideAI/self-operating-computer

1874

Victor Mimo

@vic·

langchain or llama-index?

I started in llama-index, slowly migrating over.

AI Github Contribution Analyst Update #1

- Started with a simple OpenAIAgent on llama-index with a custom BaseToolSpec to fetch all commits between a timeframe. System instructions contains a simple scoring metric.
- Quickly bumped into token limit, even for GPT4 turbo - need to index the data somehow.

1874

Victor Mimo

@vic·

Lofty goal - but I've been thinking about how to move LLMs away from Q&A agents into self-operating entities for a while now.

Interested to see this in the wild - 6.8k on GitHub! https://github.com/OthersideAI/self-operating-computer

I’ve built web bots to crawl webpages before and you really do need a human supervising it - last night I was wondering if LLMs could be good enough now with Vision APIs. Haven’t dove in, but this is a good first sign
https://www.youtube.com/watch?v=IXRkmqEYGZA

At Eth Denver I met a dev and we started talking about a better system for foundations/protocols to reward devs retroactively based on code contributions.

Simplest could be, take all commits into main for X time period - have an LLM analyze what was contributed, weight against goal of repo and rank. (1/2)

I've been looking more and more at running my own models in projects where OpenAI becomes restrictively expensive, now I can test and compare outputs easily.

https://jan.ai/

If you're looking to embed a react component from repo A to repo B, do you default to iframes?

Have you used alternatives that provide greater control? (headers, etc)

As a newbie to contributing to open-source, it feels inspiring to share an idea and have it be implemented by the authors 🚀

https://github.com/Open-Cap-Table-Coalition/Open-Cap-Format-OCF/issues/449

After coding all day and yearning to bounce ideas off another dev, I got frustrated typing on chatGPT and wanted to talk to it. I then realized what I want is Samantha from Her. Idk how to feel about this.

I got one piece of advice 7 years ago that I still think about often. It's especially relevant to founders.

"Things aren't as good or as bad as they seem"

Anyone else take a break from social media for a while to come back and have their OG friends here have like a million followers???

I don't have fomo, shut up. How do I see my damn FID anyway

I was here when @dwr.eth was free enough to do user interviews on twitter.

/Farcaster