marcopicentini

Stock is -63% since IPO

Do not take VC money.

The awesome thing for bootstrap startup is that if they go well you can get 100% of earnings and you are able to distribuite your dividends whenever you want. If you are VC-funded you have lots of constraints in this. You are almost pushed to continually re-invest your earnings every year and only get a salary.

Agree. I turned off any analytics and all 3rd party tracking. Found this you can avoid cookie banner.

I only track internal product success metrics and occasionally gather feedback from phone support. You get totally better perspective talking to users.

Analyzing data is limited because you don’t the user mind when is looking for something or clicking.

Would be very interesting to see the same documentation principles applied to other thing (not API or developer docs).

Why not also tracking eco-anxiety?

https://en.m.wikipedia.org/wiki/Eco-anxiety

What tool did you use to crete the promo video?

Why would you need to send 2000$ to an international friend?

I understand but Code LLama or CoTracker do not create content for Facebook/Instagram. Moreover, AI-generated content is not so engageable as a human-generated content. I don't think the purpose is to have more content at lower price.

The theory makes sense, but it's not clear what its complement is?

These open source AI model will create hundreds of AI startups with no competitive advantage. High competitive market and low margin.

You could use Cloud66 + Hetzner and chose Sequential deployment as a deploy strategy.

https://www.cloud66.com/

Does Cody call GPT-4 API?

Congrats! Really fast, it was just announced a few hours ago.

I would expect a VSCode plugin that make CoPilot obsolete in a matter of days.

It's just a matter of time that Microsoft will integrate it into VSCode.

Ruby on Rails, Postgres.

Ruby on Rails, Postgres.

No JavaScript, you case use Hotwire.

What are other use cases that could be made only by LLM ?

Number sorting is faster using code.

[DUPLICATED] https://news.ycombinator.com/item?id=37248494

Competition. Creating an innovative product using AI right now is too risky because of the competition. It is possible that you develop one thing and in 3 months 10 other people have developed the same thing.

Why not just say their born year instead of give a name for each generation (baby boomer, generation x,Y,Z)

What are your use case that justify a cost 40-50k/month just for training?

It’s like Cloud66 but with the GPU in the headline, isn’t? What’s difference with Cloud66?

Why not reply directly with the answer learned from paper? Currently it’s like search the more relevant paper, open the pdf and read it.

The $ amount listed on 13F is NOT the premium value but rather the value of shares the position represents (100x multiplier)

Back out the 13F SPY value: $886mm/2mm = ~$443

He owns 20k SPY puts (2mm/100) likely worth $10's of millions

It’s only for MacOSX. I expect to load the model on a Ubuntu server, not on my local dev machine.

No idea yet. Which you recommend to start? It will be hosted on a Ubuntu server (Digital Ocean, Linode etc..)

Very cool! Could you add in the readme how to install llama 2 on the same machine ?

Can I train a llama 2 model on custom data and make it expert on my data knowledge?

For example, If I give it thousands of law pages it will be a domain expert about law ?

Could you explain the use cases for you company? How do you use their data?

I liked the spaces but their price are crazy expensive. It makes not sense to spend so much for a wood desk and a chair.