As many of you, I have a few subscriptions to Claude and Codex (just me??). It's a daily occurrence to use 2-3 accounts for the same conversation (whenever the limit stops me).
What I need is something to switch the account where the usage is taken from, not the configuration.
I'm not sure providers would shoot themselves in the foot by not allowing you to buy more usage. It seems like both OAI and Anthropic could use all the subs they can get.
Tesla's taxi business isn't real. Waymo has been transparent about how their cars handle situations that require human intervention. Waymo's most important technology isn't superior sensors, it's the ability to be truly autonomous without continuous oversight, even when putting a human in the loop to make a decision. Humans never drive a Waymo remotely, because that's dangerous and stupid. Tesla won't say how their remote intervention works, but it's probably a sketchy as a Chinese Robo taxi but without the Chinese 5G network that actually is low latency.
> Roughly 20 on a given day, never more than about 50 active in a week across all markets. The genuinely unsupervised slice peaked around 25 in spring and has drifted sideways since.
Waymo:
> Roughly 500,000 paid rides per week, a figure Alphabet cited at both Q1 and Q2, so it is flat, not fresh.
OpenAI is a bit more scattered brained and spread across multiple products.
Anthropic is way more focused. That helps in a way that builds upon another: their product tends to be better and the people's perception of it as well.
But "much less"? OpenAI needs a month or two of good choices, progress & announcements, and that flips. On the other hand, Deepseek could blow both out of the water unexpectedly, ending the argument about who has the superior tech.
There's also the Elon factor. Anthropic has a good relationship with Elon (to the point where he discounted the compute sold to them by quite a lot) while OpenAI/Sama is the devil to him.
> Anthropic is way more focused. That helps in a way that builds upon another: their product tends to be better and the people's perception of it as well.
But they are still not likely to be investment-grade at IPO.
mostly because they got all those 500B hardware commitments on balance sheet.
Without them they have best coding/other(e.g. most accountants prefer claude) model on the market, with solid increasing revenue channels and loyal customer base.
I disagree with this, most normies will never use Claude code... but they WILL use ChatGPTs various modes and apps that handle a lot of things for them.
Their recent "intelligent ui" (forget what they call it), which turns the chat almost into a magazine is amazing for people when they start doing measuring or responses comeback as little graphs or diagrams showing this or that.
Claude is still not able to generate pixel images.
I agree Claude is better for the likes of you and I, but I see the opposite in non-technical people which is most people
Who says selling to "normies" is a value proposition in the first place ? if we're heading towards a K shaped economy - being mass market is a bad thing, and serving professionals and top of the curve is where there's money to be made. The only money in this market so far is in selling to megacorps and governments.
Also Google will eat them eventually in mass market part as well.
Normie consumers will probably not ever spend more than $15/mo on LLMs as they exist currently if they would spend money at all so their LTV will probably just turn into race to the bottom performance marketing
Devs and the token-consuming services they create are probably a more valuable market than the entirety of the glorified search engine use cases that the average consumer is focused on
> Deepseek could blow both out of the water unexpectedly
This is tantamount to saying that the value proposition is unknown, if a supposed trillion dollar company could get upended overnight by a startup
Where's the moat, the unassailable first mover advantage, the enterprise lock intengrations that mean nobody could realistically move you out of the way even if they had slightly better tech ? If none of these exist, investors should be scared.
DeepSeek can't come out of nowhere to match them. They don't have anywhere near enough compute for that; Liang Wenfeng said as much in his leaked comments to investors. That's the moat.
What if I asserted that the MCP economy could be worth 5x more than LLM market since tools usually end up just being a fraction of the opportunity compared to market domination.
Maybe. The problem with the MCP economy is that it's not clear yet what shape it can take. If it takes the shape of API integration solutions, then it won't be 5x. If it takes the shape of an App Store, then yes.
I think arguments for either direction are valid. On one hand, MCPs are essentially APIs used by developers. Developers will demand some form of open reusable standard, "MCP gateways" or hosts will spring up to host them, hyperscaleres will offer them, and ultimately it will be another cloud commodity.
On the other hand, some of the target users of MCPs is the consumer market rather than the developer market. An App Store model makes sense there and the provider/platform can collect the lucrative 30% cut and end consumers’ numbers makes for massive revenue opportunity. This dictates a walled garden "Agent" like Muse or Instant.
If #1 is the direction the market goes in, then more batteries-included solutions like Muse or Hermes become the next obvious step up the chain. If #2 wins, then yes having your MCP available to on a Muse App Store that Meta takes a cut of some sort would be larger.
> How could anything else be bigger than Super Intelligence? Which is supposed to be the next step after AGI, which we presumably already have?
Meh. We both do and don't have AGI, because everyone before ChatGPT thought it was obvious what it meant and it turned out we all disagreed with each other about that obvious meaning. We can't even all agree about the "artificial" part, which is the most surprising part to me, let alone "general" or "intelligence".
If you applied some thinking for a moment and not spiraled out you probably would realize the the tools that control a market are worth less to sell themselves then to utilize them.
Not sure, but in the meantime they removed the option to right click "Open in new tab" on the "New chat" button (first one in the sidebar) in chatgpt.com
> There's also the Elon factor. Anthropic has a good relationship with Elon (to the point where he discounted the compute sold to them by quite a lot) while OpenAI/Sama is the devil to him.
The Elon association could be a long-term negative.
It certainly didn't help the Republicans in this last election even with Elon donating half a billion dollars to them (post 2024 Elon appears to be, like AIPAC, more of a toxic association than any amount of money gained can overcome), and there are reasons to believe that flip on the political side will eventually matter on the business side of things.
Elon's Texas "home base" just flipped blue in a big way and there are likely to be high profile investigations into whatever it was he was doing with DOGE.
And perhaps most directly important to Anthropic, unless the winds change significantly in terms of the politics surrounding data centers (and I suspect if anything they will get increasingly hostile, not better), I wouldn't be surprised to see the Memphis city council get purged of pro-Colossus members, replaced by anti-data-center candidates when the next elections take place in Oct 2027.
> The AGENTS.md thing kind of epitomizes the whole reason I decided to quit using Claude Code.
Do I understand correctly that this decision was taken based on something that's not accurate? Claude has been reading AGENTS.md files for a while now.
> Claude has been reading AGENTS.md files for a while now.
No it hasn’t, this is a very recent addition. Previously you had to symlink CLAUDE.md to AGENTS.md, or include a @AGENTS.md in your Claude file, but by itself it would not read it
It’s the other way around. Apple don’t want to suggest offensive phases and have it look like they wants you to say those things.
The middle ground here is to not suggest anything at all. Because they’re either pushing people to say something or preventing them from saying something.
The HN crowd forgets that average people don’t know how technology works. Everyone I know would see this as a good feature.
This should be a preference then, just like in video games where profanity can either be shown or hidden in multiplayer chats. Then it makes the user aware this exists as a feature and gives them control over it.
My rule of thumb in software is: if there's enough of a dispute on how something should work, it should probably turn into a preference available to users. It's adds scope, but it respects user choice in places where it matters.
Building software long enough I've realized that the valuable part is your vision of it. The opinionated part. There's no toggle to allow p0rn in the app store, and that's good. We can argue about user choice, but undeniably even if such a toggle existed, more users would fall into addiction. Environment matters more than "self control" or "discipline".
I agree with you (on certain types of software), and I also don't see Apple as the be-all and end-all or assume that its opinions are always right. This is just my opinion based on my lived experience, both as a user and as an engineer.
What I need is something to switch the account where the usage is taken from, not the configuration.
I made something simple & transparent named Switcheroo: https://github.com/FTCHD/switcheroo
Works with other AI CLIs like Gemini, and other various CLIs like Vercel (if you're managing multiple accounts like me), Netlify, Github etc
reply