@jondot Really? What kind of libs were involved? I've revived much older TS projects without much pain. But lib selection has always been a big part of the process.
Saw someone the other day tweeting that it would be great if there was a SlackGPT that could summarize all the Slack messages for when they wake up.
And I immediately related to that. We are a team of 20, and I'm the only one in SF. So when I wake up, most of the team is already half a day in or has just wrapped up.
That means that I always spend the first 30 minutes of the day reading messages to catch-up on everything.
And tonight felt like hacking something quick.
So I created a script that:
- Reads all Slack messages from the time I go to bed
- Summarizes the conversation of each channel
- The bot sends me a message with this summary
It seems to work well, so I'll clean it up during the weekend and open source in case someone is interested in the code.
@didier_lopes ๐ฑ was just talking about this as a good first use-case at work where they take ai integration risks very seriously. Really interested in this!
So it seems we may finally have a GPT-4 level model in open source. https://t.co/uWuD5RyYU0 It's a merge of two llama 70b and since we live in the best AI timeline it's created by an anon with an avatar that looks like this:
@didier_lopes I think both has their place. Gradients always win my heart, the other one is a bit more legible at a distance. Plz do a holographic one as well :)