deepdarkforest

An agent is an llm running on a loop until it decides it should stop.

/goal is a gimmick where you run a "parent" agent on top that runs the agent on a loop until the it decides to stop, just prompting it "nope, not done yet, continue".

both amazon and uber used that spending to deliver a network effect moat/almost monopoly.

But openai's chance of a moat on model quality is dropping as we go, not increasing

lol do you not have actual stuff to prioritize for your team rather than re inventing forges? there are quite a few open source alternatives if you want something quick. Fork them. No need to re invent the wheel

compute??

Github is struggling because of compute, which comes from everyone vibecoding and triggering actions 10x more.

I can vibecode an alternative, but once i have users, who is going to secure this amount of compute? Compute+talent to manage it(devops isnt vibecoded *yet) is a moat

Its quite funny that you are building an "infinite place profile", you both worked on products used by 100s of millions of people, and yet your website is down from 45 minutes of HN traffic!

Joking, but its a very good idea. Synchronization between the physical world information and digital has been a very hard problem for decades and im sure an agentic approach can 10x the value.

Just wanted to say i think most interpretability research it's just a smoke show nowadays but this is actually the first one that i think has a very serious potential. I love that the SAE is actually constrained and not just slapped unsupervised posthoc.

How granular can you get the source data attribution? Down to individual let's say Wikipedia topics? Probably not urls?

Would be interested to see this scale to 30/70b

> You can edit it by describing changes

Even this is hard. Most people don't know what they want, and/or they don't know how to describe it/imagine it. They don't even know what a trend graph is.

They just want someone else to do the mental effort of creating a nice product. Hence iOS > android for most people. They don't want to customise basically anything other than colours.

That's why i predict Lovable/replit etc will not go mainstream. And why chatgpt will just offer you their UIs mainly. Artifacts weren't a big hit

The funny thing is that Anthropic is the only lab without an open source model

Mintlify is the best example of a product that is just nice. They don't claim to have a moat, or weird agi vibes, or whatever. It just works and it's pretty. 10m arr right there

Yes 250 gallons can save a lot of people from dying of thirst but have you considered that with that same 5kwh we can also produce 1 tiktok ai slop video of the queen boxing with mike tyson?

Never ask a woman her age, a man his salary, and an agent framework developer his long term plans

I wonder if using a local llm to override the ads would work. A finetuned one for removing ads will probably appear soon

> Codex is more hands off, I personally prefer that over claude's more hands-on approach

Agree, and it's a nice reflection of the individual companie's goals. OpenAI is about AGI, and they have insane pressure from investors to show that that is still the goal, hence codex when works they could say look it worked for 5 hours! Discarding that 90% of the time it's just pure trash.

While Anthropic/Boris is more about value now, more grounded/realistic, providing more consistent hence trustable/intuitive experience that you can steer. (Even if Dario says the opposite). The ceiling/best case scenario of a claude code session is a bit lower than Codex maybe, but less variance.

> which does pose interesting questions over nvidia's throne...

> Zebra-Llama is a family of hybrid large language models (LLMs) proposed by AMD that...

Hmmm