There's a reason why every pre-existing startup has been slapping AI all over their products - it's basically the only way to get funding at the moment.
I think this has been a phenomenon since VC became a thing, there's always been trends and fads - CEOs have had to chase them even when it doesn't make sense for the product.
I think this is the most wide-scale one I can personally recall (although I presume dotcom boom was similar), but I remember through the 2010s every database startup having to chase the same ball-game. Everyone was doing NoSQL, then everyone was doing analytics etc. Of course now all those companies have AI-native, vector-embedding, RAG, MPCs slapped all over their products.
Anecdata but from my vantage this has been true for a decade, I imagine it is only more true today, but investors were nudging to add AI to products the entire time I’ve been in tech, and failure to do that was always flagged as a risk for funding rounds. Need to show “AI/ML in the pipeline” by the next funding round or x/y/z is out…
I agree, it sounds like they're trying to expand their moat because they realised they were too late to the better type-checking party.
It's somewhat ironic that they're only just getting around to it actually. The predecessor to `uv`, `rye`, had integration for `ruff` fairly early on - but Astral have really dragged their heels on putting it into `uv` (presumably because `ruff` had already won) and even now aren't adding support until after `ty` is integrated.
Nuclear reactors are in a delicate balance, they are in constant feedback loops that means an unexpected power drop or an unexpected rise can cascade. The withdrawal of the others rods is not specifically in response to one being dropped in accidentally - it will be in response to the drop in the power output.
There's not a safety risk from doing this because the control systems are doing this to specifically stay within the stable power output.
There are some significant negative consequences from the power dropping excessively and unexpectedly. One of them is an excessive build-up of Xenon in the reactor which is a feedback loop that continues to reduce fission (known as Xenon poisoning) and is one of the underlying causes of the Chernobyl incident. Now, as far as I'm aware there are no Nuclear reactors running today that are _unsafe_ if they encounter Xenon poisoning - but it's not something you want to stumble into.
If there were a genuine safety concern related to a control rod drop then the reactor would scram and the _all_ the rods would drop and the reactor would need to be restarted.
For a personal project I’ve been piloting Spec driven development (SDD), (it’s contagious!), using Cursor and EARS statements. The strategy has been to use the frontier model to write the spec and a lessor model to write the tests and code and the frontier model to write critiquing prompts until it has nothing left to say. For my particular project there are two programs (or in human speak phases), where each program is broken up into milestones which are then comprised of a series of tasks. I experimented a lot with different models as the reviewer / spec model and the implementor model. Kimi 3 was super expensive as spec model and grok and OpenAI models always got something wrong egregiously. The Opus line of models have been the only ones to really grasp the project and I feel write great specs. Because I use cursor I settled on using groc for test routing and code implementation. I’m not sure if this is the most efficient method but I believe it’s building a large project solidly
I think this is the main one. The benchmarks from this are heavily cherry-picked, and they also widely publicised their performance for 4.5 while downplaying the fact the benchmarks were "accidentally" in their training set
The choices were clearly (far) less important than having the best model and a good product that was way ahead of the competition.
However, that's not the same as the technology choices not mattering at all. As others have mentioned the product is buggy, and it appears Anthropic are struggling to continue to release new features within the existing architecture.
That's also not to say that Anthropic made bad choices - there's always going to be a tension between building quickly to capture the market now vs moving slower and building something more maintainable in the long-term.
I think this has been a phenomenon since VC became a thing, there's always been trends and fads - CEOs have had to chase them even when it doesn't make sense for the product.
I think this is the most wide-scale one I can personally recall (although I presume dotcom boom was similar), but I remember through the 2010s every database startup having to chase the same ball-game. Everyone was doing NoSQL, then everyone was doing analytics etc. Of course now all those companies have AI-native, vector-embedding, RAG, MPCs slapped all over their products.
reply