Plenty to hate on anthropic for right now, but Ill never understand the references to output as a slot machine. It is massively a skill based tool, you CAN use it like a slot machine with "please make it work" style prompts. The variance is the difference, if you feed it great context and/or relevant sources to utilize, your odds of success increase dramatically. Slot machines, it doesn't matter how much thought you put into your pull, you will have the same odds as literally any other person pulling the lever.

It seems like there are still healthy ways to do it, I see some products sold by third party sellers that clearly are real small businesses. I google them find their real site and sometimes they offer pricing better than what they can offer after amazon fees etc.

I see that as the absolute best approach for someone like that, leverage the platform but don't allow it to be your entire online presence.

My own use case was sadly, just leveraging the platform and as all the margins tighten not only on the amazon platform itself but on shipping costs, it just gets tougher and tougher. Happy for the experiences they offered me freedom wise, but also happy to be moving on.

Well, they need to ensure AI advances, and that means advancing the podcast that will pretend that popular opinion is absurd and big tech is always right.

Erm people are not stupid, especially the investing community.

Methinks this is one of those moments where we will look back and say: Oh Sam facepalm.

> Erm people are not stupid, especially the investing community.

Yes, yes they are. It's not a community. It's a herd.

seems about right. They came across as the yes men of podcasts for tech people that want to pretend they are doing no wrong, maybe i just chose a really rough 10 minute section of a random podcast though, but not one moment did they not come across in that manner.

maybe there isnt, but as understanding grows people will understand that having an orchestration agent delegate simple work to lesser agents is significant not only for cost savings, but also for preserving context window space.

> Automate the action. For example, write a script that looks to see if the friend who borrowed money ever paid it back by a certain date. If current_date > deadline_date & money_repaid = false, then send an automated email unfriending them.

email contents: Hey, btw im not ur friend anymore. regards, not your friend

On firefox i use unhook for youtube. solves the shorts issue but im sure a lot of people would be less okay with what i prefer youtube to be, a search bar with nothing else.

Title is fantastic, had me laughing at my own ignorance to copilot's offerings before I even started the article.

I do feel like if any of the major companies could do with a rebranding it would be copilot. They are tossing that name on all of their stuff, and it just doesn't carry the weight of any of the big names even though its chatgpt models under the hood. Personally i associate it with annoying bloatware, and silently judge windows users based on if that icon is still on their tasbar.

It does a good job of when a VIP employee demands "copilot", you have to buy a bunch of different licenses for them because no one knows exactly what they want (they just want copilot, no not that one).

The opposite of what Oracle used to do (arguably successfully). Break up useful components of middleware or database servers so you have to add a litany of expensive features, similar to trying to pick a trim on the BMW website.

Sorry if this is a stupid question but shouldn't you try to clarify which product specifically they have in mind before you spend money on something that might not be the right thing?

You can try, but you will often be unsuccessful because of the sheer number of skus, all with similar and sometimes overlapping capabilities & the poor documentation on MS' part. Additionally, VIP users may not like to clarify. They saw a demo of "something" and need it yesterday...

The nice thing about being a Github shop but not really a Microsoft shop is the range of Copilot" might refer to narrows quite a bit. "Github Copilot" is a relatively coherent collection of products, and never have to be bothered with any of the other stuff.

That said I suspect this is in part an attempt at making the name so ubiquitous that people start using it the way they currently use ChatGPT. Like Kleenex or Bandaid or Google, where the name of the product is just the brand.

feels like an insult to readers to try to pretend that their revenue per month is comparable to google or apples growth when the funding is absurdly different, not to mention inflation itself.

I am very much onboard with AI within my workflow. I just don't really see a future where openai/anthropic are the absolute front runners for devs though. Maybe OpenAI does just have the better vision by targeting the general public instead, and just competing to become the next google before google can just stay google?

What is their next step to ensure local models never overtake them? If i could use opus 4.6 as a local model isntead and wrap it in someone else's cli tool, i 100% do it today. are the future model's gonna be so far beyond in capability that this sounds foolish? the top models are more than enough to keep up with my own features before i can think of more... so how do they stretch further than that?

A side note i keep thinking about, how impossible is a world where open source base models are collectively trained similar to a proof of work style pool, and then smaller companies simply spin off their own finishing touches or whatever based on that base model? am i thinking of thinks too simplistically? is this not a possibility?

Anthropic is definitely gaining ground over OpenAI in the business world. Cowork is the absolute hotness right now, and even prompted MSFT to drop their own variant yesterday

Ask anybody you know that works in Big Tech. They're all pushing hard for Claude Code adoption.

Codex and Gemini CLI seem 1-2 months behind Claude Code. They will catch up. This race will eventually be won by whoever can come up with the cheapest compute.

And that's a dangerous game because the cheaper compute gets, the more likely consumers are to self-host rather than pay a subscription.

Apple could figure out a way to neatly package it into their ecosystem.

Not really. Most people won't self host.

The general public will self-host it's built in to your next phone or laptop straight out of the box or maybe from the App Store.

I agree that that's what it would take, but compute would need to get very cheap for it to be feasible to keep models running locally. That's an awful lot of memory to have just sitting with the model running in it.

True. I was thinking more of power users. Do you think Opus level capabilities will run on your average laptop in a year? I think that's pretty far away if ever.

You can demonstrate "running" the latest open Kimi or GLM model on a top-of-the-line laptop at very low throughput (Kimi at 2 tok/s, which is slow when you account for thinking time) today, courtesy of Flash-MoE with SSD weights offload. That's not Opus-like, it's not an "average" laptop and it's not really usable for non-niche purposes due to the low throughput. But it's impressive in a way, and it does give a nice idea of what might be feasible down the line.

> how impossible is a world where open source base models are collectively trained similar to a proof of work style pool

Current multi-GPU training setups assume much higher bandwidth (and lower latency) between the GPUs than you can get with an internet connection. Even cross-datacenter training isn't really practical.

LLM training isn't embarrassingly parallel, not like crypto mining is for example. It's not like you can just add more nodes to the mix and magically get speedups. You can get a lot out of parallelism, certainly, but it's not as straightforward and requires work to fully utilize.

It's hard to train models in the open. All the big players are using lots of "dodgy" training data. Like books, video, code, destinations. If you did that in the open, the lawyers would shut you down.

Though I think these companies are wildly overvalued, I don't see LLMs as a service going away in the future. The value in OpenAI is that it provides extra compute, data access, etc. My money is on local AI becoming more of a thing, while services like OpenAI still exist for local AIs to consult with. If a local model can somehow know that it's out of it's depth on a question/prompt, it can ask an OpenAI model if it's available, but otherwise still work locally if OpenAI fails to respond or goes out of business. To me that makes a lot more sense than the future being either-or.

Models not being able to reliably know if they are out of their depth is a foundational limitation of the currently generation of models, though.

Best they can do is to somewhat reliably react to objective signals that they've failed at something (like test failures).

> What is their next step to ensure local models never overtake them?

As someone who experiments with local models a lot, I don’t see this as a threat. Running LLMs on big server hardware will always be faster and higher quality than what we can fit on our laptops.

Even in the future when there are open weight models that I can run on my laptop that match today’s Opus, I would still be using a hosted variant for most work because it will be faster, higher quality, and not make my laptop or GPU turn into a furnace every time I run a query.

If your laptop overheats when you push your GPU, you can buy purpose-built "gaming" laptops that are at least nominally intended to sustain those workloads with much better cooling. Of course, running your inference on a homelab platform deployed for that purpose, without the thermal constraints of a laptop, is also possible.

I didn't say it overheats. It gets hot and the fans blow, neither of which are enjoyable.

MacBook Pro laptops are preferred over "gaming" laptops for LLM use because they have large unified memory with high bandwidth. No gaming laptop can give you as much high-bandwidth LLM memory as a MacBook Pro or an AMD Strix Halo integrated system. The discrete gaming GPUs are optimized for gaming with relatively smaller VRAM.

You can host a website on any rackmount server for pennies compared to AWS. But people still use AWS.

The market for local models is always gonna be a small niche, primarily for the paranoid.

"The market for local models is always gonna be a small niche, primarily for the paranoid."

Have you ever heard of industrial espionage? Pr privacy regulations? Or military applications?

(Also the US military runs claude as a local model)

The goal of web hosting is to provide low latency wide availability to many users.

AI in this context has a very different goal as a tool for individual users.

You wouldn't say that hosting instances of Photoshop on servers and charging for usage is a long term viable business would you? Even if current consumer computers struggled to run Photoshop.

I don't see an issue with the comparison, I don't think it is meant to be a 1 to 1 or anything, just an illustration of how consumers are overwhelmingly lazy.

I'd take issue with the statement that it is for the paranoid, but I guess it might be a defense mechanism because of course i am interested in local models. If my new workflow is going to be dependent on 3 companies, I'd prefer if there is a light at the end of the tunnel that breaks us free.

>"But people still use AWS"

I do not, I self host. My current client is also got rid from AWS packing up nice savings as a result

honestly, the biggest reason i deep dove on proper .claude stuff, was because im a cheap ass. I saw someone mention their agents/ that delegates to cheaper models, and figure that was a way I could reign in my own overall usage, and its been true so far. Im sure im one of the very few heavy claude code users that still stubbornly sits on the pro version. It won't be forever, if i land an important contract or job, I'll pretty quickly hop to max or whatever, but for my own usage right now, im getting by.

Sure, maybe this stuff isn't crazy relevant 2 years from now, but right now? Giving your agent a clean way to navigate and delegate tasks to keep that context window clean? its 100% vital.

edit: hop to max*