Spot on. My backup local notetaker app just stopped releasing updates and became broken so I got annoyed and just whipped up my own to not lose access to my old recordings and also to just use going forward.
So much better to just add my own features and slowly customize it to suit my needs.
I have a fun feature where I can retranscribe with multiple local models and then compare transcripts and optionally label/correct them to accumulate data on which models are most accurate over time.
I can't wait until the RSI fad dies down for awhile and we can get a break from all the intellectual grandstanding...
The code they released doesn't match what's in the paper.
And the results they got are significantly narrower than the narrative heft being asked for by their word choice ("To advance beyond humanity’s accumulated knowledge", "AI’s Odyssean moment").
Don't fall for the "marketing hype" antihype. RSI is a trendy term today, and lots of people discuss it, but it's nothing new. We've basically had forms of RSI since the start of the LLM craze. Previous models have been used to improve next versions for ~3 years now. Just because humans used to guide / select / decide when and how to apply those models, doesn't make it less recursive or self or improvement. We're now seeing what happens when we remove the humans from some of the tasks, but the basics are still there. Use prev model to make next model better.
RSI is about turning that extra computation into acceleration - that is applying the compute to the (useful_work / compute) factor - this is the recursive part.
It says much more than that. It says that all custom architectures and approaches are become irrelevant if compute increases, which was proven again by the large latent models recently solving RL tasks better than any classical RL technique.
If they had something of value, as a research neolab, they would not just release it but build it into their systems and use it.
Most of these agentic swarm systems are memory and context managers/compressors. However it's clear from both the paper and the blog that everything is just natural-language-based and generated, effectively making it slop from start to finish.
Instead of describing how or what was stored by each agent, and how, in what format, how it was selected, the whole process is described as Claude-generated workflows and pipelines, which brings 0 value.
i mean marketing in some sense “feels” like fraud when presented in some very specific way, no? at some level it’s: generate interest about a thing when there didn’t exist any interest before. sometimes they generate interest about something that has no substance and that feels kind of fraudulent (this isn’t that case)
but in this case, this is product extremely boosted for super viral marketing, and this is the first time i’ve read about this boost since hearing of the product whereas everybody only really talks of its substance. feels like a magician kept a trick going for a very long time, people forgot they were watching a magic show, then the curtains come down and they reveal the deceit after (i’m being slightly melodramatic)
I'm curious which TTS model was used for the "Nora" voice. I kept noticing vocal artifacts in her speech disfluencies. Which bothers me more than it probably should...
I doubt that will be a problem. Models are different enough right now. If you can write a spec with enough detail and constraints today to get valid and comparable output from say GPT-6 Astra, Opus 5.5, DSV4F, Kimi K3, GLM 5.3, etc... Then I think there's a good chance that whatever SoTA coding LLMs everyone is using 3 years from now will also be able to implement that same spec.
Again, I think it heavily depends on if people are writing comprehensive specs with sufficient detail.
That depends on if you are really getting what you expect from the spec or you are actually relying on undefined behavior for your expected output from the spec. This is why we often pin software library versions: we may very well be relying on unintended or a bugged behavior to get out expected output.
If we fast forward say 5 years, I don't see how we don't end up in world where people (and enterprises) are more savvy with how they use LLMs. Meaning, more models, smaller models, weirder models, more specialized models, etc. And all of it running on a variety of hardware (edge devices, personal computers, on-demand cloud compute).
I don't see how NVIDIA can keep their spot as belle of the ball. If LLMs and friends are truly to become as useful and ubiquitous as everyone thinks they will, then commoditization is the only option.
We need to figure out what the real pricing is for a going concern. Right now, everyone is subsidizing and discounting to grow (or maintain) market share. The big question is whether the steady state, market derived inference pricing is above or below what we’re seeing today. I honestly don’t know. Anthropic had said that inference is profitable, but they’re clearly not yet profitable overall with training and buildouts still happening.
I’m sure someone does, but I’ve never seen anything other than vague statements like Anthropic’s “inference is profitable” comment. I suspect everyone is playing everything close to the vest because they aren’t yet public and they want to control the information flow to the street.
> I don't see how NVIDIA can keep their spot as belle of the ball.
FWIW, people were saying "ASICs will kill CUDA demand!" since the crypto mining boom. Then a few months later, CUDA found another niche application in LLM applications.
With the mounting demand for robotics, surveillance and autonomous weapons, I don't see how Nvidia couldn't keep their spot. They have their pick of the litter with hundreds of market segments, and unlike the rest of FAANG they're not afraid to branch out.
So much better to just add my own features and slowly customize it to suit my needs.
I have a fun feature where I can retranscribe with multiple local models and then compare transcripts and optionally label/correct them to accumulate data on which models are most accurate over time.