Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Claws are now a new layer on top of LLM agents

twitter.com

Post reply on HN
Print view View original

221–230 of 1001 posts

Re: Claws are now a new layer on top of LLM agents

#221

Post by CuriouslyC »

bravura wrote: post #102
ZeroGravitas wrote: post #26

So what is a "claw" exactly? An ai that you let loose on your email etc? And we run it in a container and use a local llm for "safety" but it has access to all our data and the web?

There are a few qualitative product experiences that make claw agents unique. One is that it relentlessly strives thoroughly to complete tasks without asking you to micromanage it. The second is that it has personality. The third is that it's artfully constructed so that it feels like it has infinite context. The above may sound purely circumstantial and frivolous. But together it's the first agent that many people w…

Claws read from markdown files for context, which feels nothing like infinite. That's like saying McDonalds makes high quality hamburgers.

The "relentlessness" is just a cron heartbeat to wake it up and tell it to check on things it's been working on. That forced activity leads to a lot of pointless churn. A lot of people turn the heartbeat off or way down because it's so janky.

Re: Claws are now a new layer on top of LLM agents

#222

Post by alansaber »

claytonaalves wrote: post #219

I'm impressed with how we moved from "AI is dangerous", "Skynet", "don't give AI internet access or we are doomed", "don't let AI escape" to "Hey AI, here is internet, do whatever you want".

Because even really bad autonomous automation is pretty cool. The marketing has always been aimed at the general public who know nothing

Re: Claws are now a new layer on top of LLM agents

#223

Post by wavemode »

bpicolo wrote: post #176
snovv_crash wrote: post #48

Earlier quoted context omitted.

Cron would be for a polling model. You can also have an interrupts/events model that triggers it on incoming information (eg. new email, WhatsApp, incoming bank payments etc). I still don't see a way this wouldn't end up with my bank balance being sent to somewhere I didn't want.

Don't give it write permissions? You could easily make human approval workflows for this stuff, where humans need to take any interesting action at the recommendation of the bot.

The mere act of browsing the web is "write permissions". If I visit example.com/, I've now written my password into the web server logs of that site. So the only remaining question is whether I can be tricked/coerced into doing so.

I do tend to think this risk is somewhat mitigated if you have a whitelist of allowed domains that the claw can make HTTP requests to. But I haven't seen many people doing this.

Re: Claws are now a new layer on top of LLM agents

#224

Post by alansaber »

mittermayr wrote: post #43

I wonder how long it'll take (if it hasn't already) until the messaging around this inevitably moves on to "Do not self-host this, are you crazy? This requires console commands, don't be silly! Our team of industry-veteran security professionals works on your digital safety 24/7, you would never be able to keep up with the demands of today's cybersecurity attack spectrum. Any sane person would host their claw with us…

I wonder how much the clawbase domain name would sell for, hmm

Re: Claws are now a new layer on top of LLM agents

#226

Post by tjchear »

WA wrote: post #16
cromka wrote: post #14

Earlier quoted context omitted.

But the only cheap option is 16GB basic tier Mac Mini. That's not a lot of shared memory. Proces increase bery quickly for expanded memory models.

Why though? The context window is 1 millions token max so far. That is what, a few MB of text? Sounds like I should be able to run claw on a raspberry pi.

If you’re using it with a local model then you need a lot of GPU memory to load up the model. Unified memory is great here since you can basically use almost all the RAM to load the model.

Re: Claws are now a new layer on top of LLM agents

#227

Post by singpolyma3 »

claytonaalves wrote: post #219

I'm impressed with how we moved from "AI is dangerous", "Skynet", "don't give AI internet access or we are doomed", "don't let AI escape" to "Hey AI, here is internet, do whatever you want".

I mean. The assumption that we would obviously choose to do this is what led to all that SciFi to begin with. No one ever doubted someone would make this choice.

Re: Claws are now a new layer on top of LLM agents

#228

Post by trcf23 »

kator wrote: post #73
ksynwa wrote: post #38

Why mac mini instead of something like a raspberry pi? Aren't thede claw things delegating inference to OpenAI, Antropic etc.?

Some users are moving to local models, I think, because they want to avoid the agent's cost, or they think it'll be more secure (not). The mac mini has unified memory and can dynamically allocate memory to the GPU by stealing from the general RAM pool so you can run large local LLMs without buying a massive (and expensive) GPU.

If the idea is to have a few claws instances running non stop and scrapping every bit of the web, emails, etc, it would probably cost quite a lot of money.

But if still feels safer to not have openAI access all my emails directly no?

Re: Claws are now a new layer on top of LLM agents

#229

Post by alansaber »

mycall wrote: post #187
fxj wrote: post #69

He also talks about picoclaw which even runs on $10 hardware and is a fork by sipeed, a chinese company who does IoT. https://github.com/sipeed/picoclaw another chinese coompany m5stack provides local LLMs like Qwen2.5-1.5B running on a local IoT device. https://shop.m5stack.com/products/m5stack-llm-large-language... Imagine the possibilities. Soon we will see claw-in-a-box for less than $50.

> Imagine the possibilities 1.5B models are not very bright which doesn't give me much hope for what they could "claw" or accomplish.

A 1.5b can be very good at a domain specific task like an entity extraction. An openrouter which routes to highly specialised LMs could be successful but yeah not seen it in reality myself

Re: Claws are now a new layer on top of LLM agents

#230

Post by snovv_crash »

igravious wrote: post #188
snovv_crash wrote: post #48

Earlier quoted context omitted.

Cron would be for a polling model. You can also have an interrupts/events model that triggers it on incoming information (eg. new email, WhatsApp, incoming bank payments etc). I still don't see a way this wouldn't end up with my bank balance being sent to somewhere I didn't want.

> I still don't see a way 1) don't give it access to your bank 2) if you do give it access don't give it direct access (have direct access blocked off and indirect access 2FA to something physical you control and the bot does not have access to) --- agreed or not? --- think of it like this -- if you gave a human power to drain you bank balance but put in no provision to stop them doing just that would that personal a…

What day is your rent/mortgage auto-paid? What amount? --> ask for permission to pay the same amount 30 minutes before, to a different destination account.

These things are insecure. Simply having access to the information would be sufficient to enable an attacker to construct a social engineering attack against your bank, you or someone you trust.

Post reply on HN