It may not pass muster with a judge in some backwater, but with one in the Northern District of California or a jury of their HN peers, it might. Laws are what we make them.
Depends on what you're selling. Patek Phillipe, Vacheron watches? Never mind business hours, it's "By Appointment only". A USB-c to USB-a adapter that I need tomorrow? Yeah you better be open buddy, or prime will.
Exactly, as I've said upthread, DeepSeek + Together.ai, Mistral, Meta + Grok...there are too many opensource + infra teams for this "sealed model" strategy to ever work.
Cursor uses both Anthropic and OpenAI and benefits from both. Would <FoundationalProvider> be able to build CursorKiller and be competitive with Cursor without <OtherFoundationalProvider>? Without a multiple of current model performance, which don't think is a stretch to say is unlikely, I don't think they can.
I just saw a billboard sign for “deepseek on together.ai” on the 101 a couple days ago, and thought “that’s a very good idea”. As long as there are infra players like together.ai and truly open source crack research teams like DeepSeek and Meta (no one can monopolize research(ers)) I don’t think this “gated model” thesis holds. Which means that OpenAI and Anthropic are competing with every other token output wrapper, and somewhat poorly as they aren’t able to offer competing models where theirs may be lacking. Cursor defines this dilemma for the foundational providers.
Resources still need to be allocated. Even by the AI itself. At some point a group of neurons will “argue” for more resources, and then within that group a subgroup will, and so forth. Hence the paperclip maximizes…
Was nodding along until "...consolidating databases into a single underpowered behemoth...". Nothing worse than race conditions and stale data between multiple disparate datastores. I'd rather have one database and more ram for as long as possible, tyvm. Some of the latest nights I've had debugging were at companies with mission critical data spread across multiple databases. I know it's orthogonal to the point, but this one really didn't land with me.
Exactly. There was a blog post a couple years ago called “The Tyranny of the marginal user” that states this principle succinctly. If it’s anymore effort than a thumb swipe, you’re losing users in a hurry.
How many “executives” does meta need vs trad media companies in the 1900s - 2010. Those companies and by extension the execs that ran them have been entirely replaced by automation and AI. That’s just one example. I could name dozens of industries affected in this way. If you want to make a lot of money, look for something that’s really hard that only an “executive” does. Here’s another one: allocating capital. Investing is now increasingly run by ETFs and automated managers, not the corporate raiders of the 80s and 90s. Those people still may have monetary wealth, but not the means of production.
I disagree. In fact they’ll be the FIRST to be replaced, as that is the richest target for replacement. Think about it. You wouldn’t build a robot (R+D costs) to do the job of someone cheap unless you can build many many robots and amortize the cost over millions.
Soldiers for instance are EXTREMELY expensive; so in modern theaters they’re being replaced with drones and missiles.
And executives are being replaced with automation in that there are less and less of them and the ones that remain have much higher leverage.
What they really meant by this is: “Competing products [at peer companies]”.
This is still in effect, however the primary competing products at peer rivals is now “foundational ai”, and they need far less people and mostly people of a very specific skill sets of ML research, Inference Scaling, Hardware R&D, AI prompt engineering research. Also the cost to play is 10 billion minimum.
So Salesforce will perform rent seeking as long as their moat allows rather than gamble on foundational ai.
Have an upvote! I've been trying it out, it's quite nice. What I like about this vs CoPilot and Cursor is that I feel like (especially with CoPilot) I'm always "racing" the editor. Also Cursor conflicts with some longstanding keybindings I have, vs this which is just the terminal. Having worked on a similar system before, I know it's difficult to implement some of these things, but I am concerned about security. For instance, how well does it handle sensitive files like dot.env or gitignored files. At some point an audit, given that you're closed source would go a long way.
I could make the same argument for every form of entertainment. Sure, we should all be hunter gatherers and only have sex to procreate. It's not realistic.
Because of switching costs. If you start a new thing this is definitely the case. It’s often said that a new product (startup), can’t be a marginal improvement; it needs to be 10x better. 95 percentile is not 10x
> Uber is objectively worse for every single party involved.
Wrong on at least one count. I've never been refused service while black from Uber. The taxi industry was brought out of the dark ages of discrimination by Uber et al. Taxis (around the world) have tried to rip me off almost half the time I've used them, with no accountability.
He has delivered on many things, and not delivered on many other ideas. It’s just that the things he didn’t deliver ( the vision ) helped him sell stocks as well as cars, via fame.
Love the idea and isometric style! I played with it, I think the dragging needs some work...
Edit:
* Enter should leave the editing modal with the updated title reflected on the "node"
* I want to drag my "nodes" around but it's not clear how to do that.
* Can't edit or delete the nodes ( maybe with an authenticated account? )
* Wishlist: right click context menu to do the above
* Wishlist: should be able to drag the scene without needing to select the hand icon