Big risk. Especially if you have memory enabled or have not toggled off the “ok to train on my data” toggle.

I do the same but a co-founder usually has different view or opinion. LLMs (in my experience) are way to agreeable and just tend to agree and not push back much.

I've seen that too. My master prompt specifically asks to challenge the question/request with critical thinking and play devils advocate

Care to share? Thanks!

can't share the PRD context obviously :) but here's the master prompt: https://gist.github.com/just-digital/d695c4956edc9e76d9d4860...

Of course I was referring to the master prompt :)

Thanks for sharing! Very helpful to get a better discussion

Hi thank you! more than 80k demos in the same minute! try now!

Will OpenAI models be available in Bedrock?

It says

> Non-API products may be served on any cloud provider.

I am not sure if Bedrock counts. There are 2 OpenAI models already there: https://aws.amazon.com/blogs/aws/openai-open-weight-models-n...

I would imagine they couldn't offer models through Bedrock. I think this means training and traditional computing workloads for their products (such as the workspaces for Codex cloud)

Does Argentina have a non-redacted version of the Epstein files or something?

It is even worse. Argentina can ask China for help.

There's no way the administration in power thinks that's worse.

Tricky. Yes you'd think that playing China and America against eachother would be a smart move.

But Americans have a proven track record of killing South American leaders and invading their countries.

Perhaps one day, god willing, the Chinese will be in a position to send aircraft carriers to Panama.

Ally for future war w/ venezuela?

Venezuela is pretty far away from Argentina isn't it?

“He better watch out”. What times are we living in.

I run chromium on my server and render the PDF from there using puppeteer.

Looks great, congrats on the launch! Which STT model are you using on device?

Grok is my least favorite model. Interesting.

Is it purely chat based (ad-hoc) or users can make dashboard like “cards” for common queries they have? (Imagine the stripe dashboard MRR, today’s revenue, failed charges, etc)

Purely chat based for now, though it has crossed our mind that replaying our generated queries might be useful.

We haven't figured out what form that might take yet.