Anthropic's goal is to commoditize intelligence. People who use their brains / intelligence for competitive advantage might not want to contribute training data for that goal.
My wife's startup made the mistake of building her internal operations around Claude Team.
Then she hired a VA in the Philippines. Anthropic promptly banned her account without warning once the VA connected to the account. It took her weeks to get her account reinstated, at which point she had already moved on to OpenAI.
How many big tech companies let you talk to a human to get support.
Automation is wonderful to cut cost for them but for the users being unable to get support is a horrible experience.
But you cannot go elsewhere because they are the only player in town.
How can small companies with 1000x less money able to provide live support, but if you pay 20, 100, 200 dollars for a subscription you dont have a phone number to call ?
I work with music streaming, I don't really know how to judge what is low-ticket vs high-ticket. However it is a medium-to-high margin business.
Our free tier is time-limited but we still look at all tickets even from non-paying customers (in the hopes of converting them). A 1-hour intervention from a customer rep can result in a multi-year paying customer.
Vanguard does get some support right as they tier their support based on how much you're worth. Businesses need to focus on Lifetime Value of their customers and realize that some of their marketing budget would be better spent in support.
>Automation is wonderful to cut cost for them but for the users being unable to get support is a horrible experience. But you cannot go elsewhere because they are the only player in town.
I can't tell you how many times I've experienced this with comcast. The last time I had to deal with it, was when I bought a new cable modem. I call in to provision it, the automated system assumes I have one of their modems and fails. For some reason I can't get technical support on the line and finally I resort to yelling 'cancel my account' over and over again until I finally get someone on the phone.
The guy was able to solve the issue in 5 minutes flat. The problem with automation is it's only ever going to be able to handle the 'happy path'
It feels so degrading talking to the bot. Last time I did it I was trying to upgrade my service to take advantage of a 2.5GbE modem I bought and I almost said screw it because it was so frustrating with the long pauses after everything I said!
The company solves the happy path every time. Your problem is they are solving their happy path which is profit optimization. The system is not poorly designed, it is working as intended.
The solution here is removing corporate monopolies and political power.
> How can small companies with 1000x less money able to provide live support, but if you pay 20, 100, 200 dollars for a subscription you dont have a phone number to call ?
They spend the money which can drastically cut into their profits.
Part of it is scale. If you have a small number of clients/users it is possible to provide that support. If you have millions or billions of users then you can't scale the support to handle the support requests, so some form of automation becomes inevitable.
If you scale up customers, you scale up support. If you don't want to serve new customers, tell them to take their business elsewhere. But if you want to be a big boy, you need to play like you are one.
If your maximum addressable market is “the whole economy,” as seen in SpaceX filings, then a city-sized call centre (distributed, of course) really is ‘t that much of an ask.
If you think you hit a limit where you can't support any new customer anymore, you just tell them you have no capacity at the moment. Like any normal practice does.
If firms had a base degree of customer support they were expected to provide, they would still exist. They would just not be as profitable, but customers would be better off.
I seem to remember there was a time when S/W was also designed with the aim to be easy to use, so that the need for support was reduced. It feels like the lesson learned was to keep costs low, not to ensure users were ok.
So you overextend, such that the quality of your support suffers? You can just call it what it is: greed. Have you considered that maybe a company shouldn’t have millions or billions of users? That’s a lot of eggs to put in one basket.
If you could be profitable with 50 customers providing excellent support, you can be MORE profitable spreading that excellent support across a larger customer base.
Just because none of the big corpos choose to do this does not mean it isn't possible. It's JUST greed.
A lot of companies seem to want to lock in to one solution or the other - like picking Oracle or SQL Server. The landscape is far too unsettled for that imo.
This is why I re-did the AI operating system I originally placed in Claude. For my 2.0 version I pulled it into OpenClaw (then eventually migrated to Hermes). All our work can be preserved after we change the model, even at a moment's notice or temporarily.
Right now we're using OpenAI's models by default since (unlike Anthropic) will allow us to use our pro subscription rather than token metering, but I've already had the joy of being able to change it to Kimi K3 (via OpenRouter) for an hour to try it out, and there were zero hiccups.
we have 7 coworkers we have been trying to re-instate for nearly 4 months now. All using the same google workspace sso, so there really was no special reason to ban them...
I had my GMail account locked for 1-2 years because I accessed it from my parents house (in the same country but in a different county) while on holiday. That was because they detected the account being used from a different IP address.
How do you know there wasn't account sharing though?
I've seen some amazingly dodgy stuff when hiring people from south east Asia, sharing a paid account with friends worth a months rent there seems milquetoast in comparison.
This is such a tough problem. Anthropic would need access to some kind of technology that could, like, intelligently handle unforeseen circumstances and nuances. Yeah, that’s definitely not something we should expect of them.
Just use any product going all in on AI hype. In 5 minutes you will see annoying bugs, server is down, non-sensical press releases, confusing UI. This is GitHub, this is Cursor, this is Anthropic, this is Google, this is all of them.
Including most of vide coded apps one sees, even from people who they'd trust before.
Anecdotal example, I downloaded a new alerting app recently from an indie dev who had a small following back in the day in iOS space. It asked for a subscription, like $20/year.
I thought, let me try this the (final version, from Mac App Store) app first. Well, it's a barely-there vibecoded shit. There's a bare-bones list, everything looks like my nephew designed it, the macOS "app" is a iPhone-size view of the iOS one, it has a bug that if you click on it it opens multiple duplicates of the same list view for no reason that you have to manually close, and in general it barely works.
Not to mention if their tools were so clearly useful they wouldn’t spend so much time making UI updates designed to force, trick, or confuse me into using their tool when it wasn’t my intention. SaaS companies with assistant integrations are the worst about this (looking at you, HubSpot)
Agreed. I think LLMs are best used as pair programmers or typists for users who already know what they’re doing. Or as tutors for users who want to learn.
Vibe coding is mostly garbage. But it can be useful for creating instant, disposable prototypes to investigate an idea or design direction.
Well a VA would need access to her account, just like how they would need email access etc.
But Anthropic has clearly picked the enterprise side of things, small teams and startups without millions of dollars in token budgets are irrelevant to them. I’m surprised they even got their account back to be honest.
And yet enterprises will be the first to move to on-prem LLMs as soon as they become feasible. The next generation of TPU chips already promises 5x efficiency and enough RAM to run a 1TB+ model, and Kimi K3 is about as good as Fable for a lot of tasks, so we'll be there much sooner than anyone anticipated.
What? I get so much joy out of learning the details of careers of people in different industries than me. I had an hourlong conversation with someone the other day who is in the high-end rug business…where he sources from, how he deals with difficult clientele, how he gets new leads, what it’s like visiting the remote villages where the rugs are made, etc. And another one with a hedge fund quant, and a separate one with a professional dancer. These are some of my favorite conversations to have with people.
They invented a dumb fix and complained that it wasn't good. Or, since we're being artistic in this thread: pulled a straw man out of their ass and complained that it smelled foul.
I did the same with cancer/mortality to demonstrate the same trick in a setting where its flaws were more obvious. It's true that I said the quiet part out loud in a way that the post I was mocking did not, but the quiet part is especially important to debunk so I make no apology for doing so.
I don’t see how you can make this claim when the budget balance at universities is extremely tilted towards stem. The budget for the kind of programs you seem to be complaining about come out to percents. Hard to say universities are “more interested” in very very small parts of their budget. What is happening now though is the federal government holds back billions of dollars in medical research grants, punishing, in some cases even killing patients, over “ideological” issues.
Ok, there probably are folks who can't differentiate between the Physics department and the Sociology department. Or who can tell the difference but who attack both anyway.
The problem for science is those folks. Right? The universities are actually teaching and advancing science. They're not the problem. The Pol Pot types are the problem. We know this very well from history.
While i know better to argue with conservatives, I would just like to remind you that if the political tides ever turn, and somehow Democrats manage to take back control of the government, its going to be actually bad news for people like you. There are already several projects on the way compiling political associations (through buying advertising targeting data), and when it becomes socially acceptable to discriminate against anyone with conservative views, you are going to be in a world of trouble.
This comes up every time this conversation occurs.
Yes, PG can theoretically handle just about anything with the right configuration, schema, architecture, etc.
Finding that right configuration is not trivial. Even dedicated frameworks like Graphile struggle with it.
My startup had the exact same struggles with PG and did the same migration to BullMQ bc we were sick of fiddling with it instead of solving business problems. We are very glad we migrated off of PG for our work queues.
The issue is that "83 per second" is multiple orders of magnitude off the expected level of performance on any RDBMS running on anything resembling modern hardware.
I haven't worked with Graphile but this just doesn't pass the sniff test unless those 83 jobs per second are somehow translating into thousands of write transactions per second.
Their documentation has a performance section with a benchmark that claims to process 10k jobs per second on a pretty modest machine, as an indication.
> The issue is that "83 per second" is multiple orders of magnitude off the expected level of performance on any RDBMS running on anything resembling modern hardware.
This is just not true, there are so many scenarios where 83/sec would be the limit. That number by itself is almost meaningless, similar to benchmarks which also make a bunch of assumptions about workloads and runtime environments.
As a simple example if your queue has a large backlog, you have a large worker fleet aggressively pulling work to minimize latency, your payloads are large, you have not optimized indexing, and/or you have many jobs scheduled for the future, every acquire can be an expensive table scan.
(This is a specific example because this is one of many failure scenarios I’ve encountered with Graphile that can cause your DB to meltdown. The same workload in Redis barely causes a blip in Redis CPU, without having to fiddle with indexes and auto vacuuming and worker backoffs.)
There are good arguments for it, but it's also not a coincidence that they happen to align with Google's business objectives. Ex it's hard to issue a TLS cert without notifying Google of it.
Google also knows about every domain name that gets renewed or registered... How does knowing a website has tls help in any meaningful way that would detract from society as a whole?
The certificate transparency log lets everyone know which domains are active as the certificates are getting renewed, likely more often than the domain itself, and also which sub-domains are active if those are not secured using a wild-card certificate.
Not just Google: AI bots could use the information to look for juicy new data to scrape and ingest.
Probably not a significant thing, the information can be derived in other ways too if someone wants to track these things, but it is a thing.
Not at all IMO, unless you are really paranoid about Google & friends. I was just saying that what was being questioned does (or could) benefit them a tiny bit.
That will not work with many of the world's most important documents because of information density. For example, dense tables or tables with lots of row/col spans, or complex forms with checkboxess, complex real-world formatting and features like strikethroughs, etc.
To solve this generally you need to chunk not by page, but by semantic chunks that don't exceed the information density threshold of the model, given the task.
This is not a trivial problem at all. And sometimes there is no naive way to chunk documents so that every element can fit within the information density limit. A really simple example is a table that spans hundreds pages. Solving that generally is an open problem.
> How many degrees of freedom do you real need to represent API cost.
The amount that your users care about.
At a large enough scale, users will care about the cost differences between extraction and classification (very different!) and finding the right spot on the accuracy-latency curve for their use case.
Exactly correct! We've had users migrate over from other providers because our granular pricing enabled new use cases that weren't feasible to do before.
One interesting thing we've learned is, most production pipelines often end up using a combination of the two (e.g. cheap classification and splitting, paired with performance extraction).
reply