It won't be sustainable in the long term, but it doesn't need to be. It's about throwing everything at the wall now and hoping some of it sticks before the OpenAI and Anthropic IPOs take off. Then just like Google Search you'll start to notice the more intensive AI features are randomly missing.
Given how bad the spam filtering has gotten, the way they give spam notifications from Google Calendar a free pass, the current mess that is "Priority Inbox", I have my doubts they can be trusted to correctly filter my new "AI Inbox" to catch what really matters.
This lets high-stakes items — like a bill due tomorrow or a dentist reminder — rise to the top.
I really hope it hadn't just been trained on US content because their so called decades old capability to have flight reservations appear in Google Calendar and Wallet sourced from GMail has never worked for me.
probably not far off it, and to be honest, that's actually better than 99% of spam filtering that exists.
I think it's more likely a "confidence" between 0-1, and then you can customise what is an acceptable threshold per account. Business accounts needing confidence 0.7 and free accounts needing confidence 0.6.
That's how I'd think about it at least. Boolean is slightly too lossy.
It's a surprisingly hard problem. I filter / rank my own emails this way and the prompt has grown to like 2,000 words and it's all very specific to exactly what's important to me. The latest Gemini Flash works well but older mini models struggle.
This is especially bizarre because one thing LLMs have been better at that practically all the developers I have ever worked with is writing good commit messages. The fact they didn't make use of this here when everything else in Claude Code seems vibe-coded these days is funny to me.
Yeah... I had a fairly in-depth conversation with Claude a couple of days ago about Claude Code and the way it works, and usage limits, and comparison to how other AI coding tools work, and the extremely blunt advice from Claude was that Claude Code was not suitable for serious software development due to usage limits! (props to Anthropic for not sugar coating it!)
Maybe on the Max 20x plan it becomes viable, and no doubt on the Boris Cherny unlimited usage plan it does, but it seems that without very aggressive non-stop context pruning you will rapidly hit limits and the 5-hour timeout even working with a single session, let alone 5 Claude Code sessions and another 5-10 web ones!
The key to this is the way that Claude Code (the local part) works and interacts with Claude AI (the actual model, running in the cloud). Basically Claude Code maintains the context, comprising mostly of the session history, contents of source files it has accessed, and the read/write/edit tools (based on Node.js) it is providing for Claude AI. This entire context, including all files that have been read, and the tools definitions, are sent to Claude AI (eating into your token usage limit) with EVERY request, so once Claude Code has accessed a few source files then the content of those files will "silently" be sent as part of every subsequent request, regardless of what it is. Claude gave me an example of where with 3 smallish files open (a few thousand lines of code), then within 5 requests the token usage might be 80,000 or so, vs the 40,000 limit of the Pro plan or 200,000 limit of the Max 5x plan. Once you hit limit then you have to wait 5 hours for a usage reset, so without Cherny's infinite usage limit this becomes a game of hurry up and wait (make 5 requests, then wait 5 hours and make 5 more).
You can restrict what source files Claude Code has access to, to try to manage context size (e.g. in a C++ project, let it access all the .h module definition files, but block all the .cpp ones) as well as manually inspecting the context all the time to see what is being sent that can be removed. I believe there is some automatic context compaction happening periodically too, but apparently not enough to prevent many/most people hitting usage time outs when working on larger projects.
Not relevant here, but Claude also explained how Cursor manages to provide fast/cheap autocomplete using it's own models by building a vector index of the code base to only pull relevant chunks of code into the context.
Old thread, but I just realized this domain was registered on 2025-12-27 11:49:33 UTC. Some DNS servers (e.g. nextDNS) block recently created domains, to curb spam and malware.
Also, for some reason Google Safe Browser thinks some pages on this site are unsafe [1]. It may well be a false positive, but it further contributes to the blocking. So it wasn't the slashdot effect after all.
Even with compression on, running most apps like a web browser over x11 forwarding, is slow to the point of almost being unusuable.
However running web apps over forwarding is pretty decent. VS Code and pgAdmin have desktop like performance running in the browser SSH port forwarded from a remote server.
UIs used to be more responsive on slower hardware, if they took longer then the human reaction time, it was considered unacceptable.
Somewhere along the line we gave up and instead spend our time making skeleton loading animations as enticing as possible to try and stop the user from leaving rather then speeding things up.
It's truly an amazing sight, our payroll system was all text based screens. I had a question and the clerk ripped through like 10 screens to get the information I needed, we're talking 200ms human reaction speed through each screen.
I also worked with a mythical 10x developer and he knew all the Visual Studio keyboard shortcuts. It was just like watching that payroll clerk (well, almost, we had under-specced machines and Visual Studio got very slow and bloated post v2008), I don't think I ever saw him touch the mouse.
reply