Super intrigued by this - large scale automation using LLMs is quite annoying due to deprecation cycles of models from frontier labs and cost of running your own being prohibitive when you have a blend of them.
I wish people could read through the puffery more easily - this company does nothing except prompt frontier models and act as if they are discovering or inventing capabilities. Their previous announcement completely misappropriated the concept of autonomy with human in the loop regardless. The harness they alleged to have created here is a commodity and as byproduct of more capable and frontier models, there is not really anything they have contributed. Otherwise you wouldn't just see monotonically increasing scores with better models
Like Airtable - they probably started seeing revenue growth/acceleration plummeting due to vibecoding being able to replace the product and are pulling plug early instead of letting it grind out to 0.
This would be incredible for innovation if we can bring this back domestically. It's very hard given lack of raw materials and the supply chain around it often also requires robust network but hopefully this spurs something.
This is something that takes real expertise though, kind of a moonshot for an inexperienced team unfortunately. Would be cool if had mission driven backers otherwise will just end up being hobbyist material. Relying on defense (bubble?) gives at least some breathing room as not actually competing against china for better prices.
Can someone give me an example where this is truly and uniquely useful? I've seen so many of these things and I can't tell if part of it is like mostly for enabling nontechnical folks to do more things or if it's some unlock and additive value. At end of day beneath it all things are just prompts, and then you can provide tools and context, but even then that's not even something I've found that useful to keep because context windows are still limited and bias propagations often need to be constantly corrected, especially when digital artifacts are not perfect representations of the real world.
kuzudb was sunsetted. I think for graph workloads duckdb is best bet, even though this is just a community plugin the upstream duckdb has really robust community doing a lot of heavy lifting and updates.
Trying to ascribe causality on macro/global level when the interesting tidbits are certainly on the regionalized scale seems mostly useless, although somewhat there is irony in again a man writing about why he thinks women having less kids than just going around and getting actual data points that aren't capture by societal level statistics.
I think the undeniable trend is caused by economics especially in the western world. It's more expensive to live than ever, and kids are very expensive when you factor in cost of living and lifestyle adjustments.
This sort of essay is farcical in ascribing economic stability as "status" - which of course has a negative connotation of seeking status, but suppose thats on track for a man writing an essay about this. Women( and men) need to work now because single incomes are not enough, and women biologically have larger burden in child rearing regardless. Economic incentives for child rearing are a joke, much like most policy that ends up more as a panacea than a solution - because none of it targets the other huge correlation we see, the growing level of inequality of wealth where the top n% is accumulating more and more. One need not look at number of lawsuits regarding women getting fired or laid off from work for needing/taking maternal leave. In the US you often need to be tenured for a year to qualify for it. These are all economic issues at it's core, and the tautology to it being status seeking is an damaging one in and of itsef.
despite having multibillion dollar valuation and a real product and service doordashes tech blogs have always been surprisingly simplistic and borderline elementary or in this case, kind of just slop. I remember reading their early data science ones and they were all sort of comically limited or discussing tradeoffs between outdated methods.
The whole point of agentic tool calling is for there to be runtime decisions made by the agent based on the outcome of a tool call. Anything that can be serialized like this could effectively be rewritten as a single tool call which this seems (via the example) to just be a tautology of for the most part. Otherwise you would have combinatorial explosion of all the paths if there is actual logic branching.
fwiw, this could be a bug but the submitters level of arrogance places this rather high on the dunning-kruger side of things. There are multiple other plausible explanations, but this person is probably vibe coder who believes anything an llm says (including explaining its own hallucinations)
The massive information loss from officials who do not understand tech communicating about this has lead to these outrageous claims. I am super interested into what actually the researchers found, as the summarized headline seems highly unlikely, or at least in need of a lot of disclaimers. Especially given that mythos seemed to be more about scaling test time compute and orchestration rather than leap in intelligence.
What I've finally come to understand is that there is a large amount of people who are now able to write and use software through claude and coding agents. Those people have different needs than more traditional software engineers who have more knowledge because even best llms often need steering, correction, and refactoring suggestions when iterating on code and it's fine to let it lose context because exactly like you said, you tell it to read file and then have to regurgitate the understanding so you can correct or validate it before continuing.
For those where the code is almost entirely a black box and cannot easily recover when something goes wrong. They are much more keen on this context management and planning because recovering from derailments is much harder (and takes longer) because its often a conversation with llm to try to recover to where they were before.
probably all AI slop but I find it hilarious in the blog post they actually posture like they would know how to fine tune a model to sound like them given that what they actually did is something that you could one shot with claude if you knew what you were doing.
This article and ones like it are nonsense, there is really no control group as they try to make like an absolute comparison of things instead of relative to in office work. You don't just take pros and cons with respect to a vacuum. I see no mention of all the pros and cons of in office work of which there are tons obviously.
factors are not invariant under different economic levels though. For example - if you didn't need to work you can spend all the time in the world with your kids. Someone who has no money and relatively low mobility and has to work night shifts at a factory does not even have the option to consider staying home like that.
can someone tell me the actual market size for technophiles who I think are only people who use this stuff? I get lost in fact that people can't really understand that end of day this is just llm calls with like sqlite facade. I see the value is the convenience only of not having to set stuff up, but everyone else claims it does all this extra stuff that is not trivial to reproduce yourself.
they kind of do.. gcp has their lambda equivalent which i believe comes with chromium preinstalled, its how major search tools like jina work, sure thre problaby somethign about session management that they probably neuter to prevent abuse though
Given it was made by cognition (team behind devin flop) who now just got to wait out until claude and gpt5 basically do all of the work for them - not very. When you read about it, the framework is highly subjective. Which very quickly becomes a problem because its based on heuristics that probably change a bunch with a better code model.
In what world or frame of reference would doing TDD have "little" bearing on output quality? If you build a system around satisfying some set of requirements it seems logical that output quality would have pretty heavy correlation.
This is talked about all the time in the US as a huge impending problem with articles about all the time. You have segment of population with higher education and higher wealth that are having fewer children and later in life as it takes more time to get secured financially and also children are very expensive in terms of maintaining what used to be a middle class life style.
why is "solving" the issue somehow the bar? software engineering has more practices rooted in psychology than engineering, its a moving and ambiguous target. Using conventional commits gives you a framework and mechanism that undoubtedly improves contribution semantics.
From my own experience in big tech and various OSS things, the crux of this piece rests on this whole scope not being focal point misses that in general the other practice is making sure PRs have limited scope in general so it's marginalized, i.e. reducing blast radius or too big of a change at once by design is implied.
You are missing the point. Of course people want to live in one of the most vibrant cities in the world. People also want a vacation home at the beach or in the mountains thats private and beautiful and easy to get to. Except if we built giant monstrosities and condos in the hamptons and make all ski homes tenement housing it will be much less desirable to go to them. No ones asking to make more apartments and housing in rust belt cities.
I am a New Yorker, people want more housing but there is still NIMBYism because they want to preserve the charm, and I'm mostly only talking about manhattan. While people are not fans of the low density luxury skyscrapers popping up in places, I've not seen people who currently live in the place think we should add massive housing blocs carte blanche. Sure there few scattered places for a few projects but not like advocating to tear down to build bigger. That mentality comes from people who are definitely not new yorkers or live in fringes.
That only seems plausible if whatever corpse of xAI is around is giving them engineering time. I don't know if they hired a bunch of ex frontier lab staff but its unlikely they have the technical capability to train their own frontier models especially the pretraining. Because the thing is if its not competitive with claude/codex it will be panned.