>... but as there's no complete feedback loop, it still would require a lot of human effort.
Not for long. Picture this: a robot receives instructions on what to physically solder in order to complete the desired modification task.
However, before it can send an image back to the vision-aware LLM guiding it, the PCB lights on fire along with the robot because said LLM confidently gave the wrong instructions.
Then, the robotic fire brigade shows up and mostly walks into walls unable to navigate anywhere useful.
I'm already having lots of success letting the agent loose on the arduino or rpi and figuring out all the annoying i2c bits and having me try different pinout and wiring combos until it works. Even with a human in the loop agents are useful right now for electronics. On one occasion I did give it a camera feed so it could check for itself if the LEDs were doing as expected.
Morality aside: To be fair, having ran many legitimate businesses and knowing few people that did the opposite, I must admit that the difficulty level in running "elaborated grey activity" is actually quite complex, people have this belief that it's actually "easier" but I doubt it's the case, many many more guardrails (accounts, anonymity, money where it goes, how can it be sustained...)
That's right. That shit is hard. But it's the only possible niche for someone who's an outsider and doesn't have a Valley network: do something legit companies can't do for regulatory reasons. Otherwise, well-funded companies with deeply networked founders where both funding and actual sales are done between people who knew each other from school or are relatives, they will just eat your market and will never even notice you.
I work with a company that outsources to a foreign IT firm. They are slow, expensive, and the quality of work is poor. They often hire subcontractors that are kids just out of school, now doing everything with AI. I've seen their prompts and often they are little more than "fix the tests." They charge $200+ hour or more for this in some cases. Insane.
And here's the trick: they use insiders to negotiate these deals, give massive kickbacks, and sometimes literally take over and destroy their victim (and i literally heard this term for a 'client' in outsourcer chats - 'victim')
Yep. I'm pretty sure the people who negotiated this "deal" did it for their own self interest, since they later went to work for the contractor directly. Too bad. We could've hired some solid independent developers.
I wasn't judging nor disputing that, just think it's a sad commentary on the current state of the world when outsourcing is considered as dirty as scam and cam girl operations.
There's a major difference: scam and cam girls work and make money, outsourcing these days... not so much. Even top outsourcers who have armies of lawyers and can afford to play harder than others, are in a sorry state: EPAM stock is 5x down from peak while rest of the stock market is 2x up from same point.
>I'm a pre-launch sole proprietor who put all my personal savings into this startup. This bill would financially destroy me for usage that generated zero business value.
That's terrifying.
If I'm not mistaken even AWS tends to forgive instances like this, so here's hoping Cloudflare has a similar disposition.
Pomodoro and todo list apps are so yesterday. Now I can have my graphics card observe me as an ever-vigilant guardian of productivity.
That might sound sarcastic, but moving context between prompts and just keeping the gears turning often isn't really that cognitively engaging these days. Thus, attention suffers.
Using Claude 4.6 [Extended thinking] exclusively via the web interface: No, not really.
My use case is having Claude tear through an extremely complex Kubernetes setup, reviewing code and drafting plans.
Despite the near-instant answers, it still manages to do this effectively at a speed I can't even hope to keep up with as a human. It's reconciling concerns that easily span dozens of dimensions with each problem I give it.
The trade-off here is that you sometimes see the model make subtle errors in thinking, but they're easily recognized and corrected for when called out. I've also noticed the model will make a statement and then correct itself mid-stream, which sometimes muddles my job of reviewing its output.
Compared to competing top-tier models taking anywhere from 5-40 minutes for a sometimes impeccably-reasoned answer, there's no comparison velocity-wise. The real win is the speed at which Claude troubleshoots, though. Near-instant turns really wins here.
It's tempting to directly assume speed is proportional to quality, but we really don't know what's going on at any given provider's back-end serving configuration, nor the internal model routing configuration.
>... but I found out they were planning to mount teleoperated weapons on the robotic platforms for a demo. I’m not willing to go there, so I resigned without another offer.
There's a ton of anti-drone defense startups currently, but how does one defend against armed bipedal robots? They're a lot heavier than most drones, so presumably they can be decked out with various types of shielding.
As far as I'm aware, phased plasma rifles in the 40-watt range don't yet exist.
I'm putting my money on miniature RC Incom T-47s complete with tow cables.
It should be required that as soon as firmware detects that its bipedal host has, in fact—had its legs tied up by a snow speeder—that it must proceed with a claymation-style display of panic before falling onto its face.
Now you've got me curious about what the Bible has to say about crustaceans, and if anyone's already created a religion around Rust [the programming language].
Not for long. Picture this: a robot receives instructions on what to physically solder in order to complete the desired modification task.
However, before it can send an image back to the vision-aware LLM guiding it, the PCB lights on fire along with the robot because said LLM confidently gave the wrong instructions.
Then, the robotic fire brigade shows up and mostly walks into walls unable to navigate anywhere useful.
The future is bright.
reply