I very intentionally have all memory turned off for chat. The amount of times I want to discuss an approach for it to pull memory out and have that steer the decision making is so obnoxious
When casually a friend asks you to ask your Claude something about topic you chatted about earlier and then Claude brings back a secret you didn't want anyone to know yet.
Yes you can set up a project and then ask a question, but this is tedious.
codex (their app) is pretty good and lots of banked resets, luna is cost effective and Astra seems better than Fable for many tasks. Most important one is less refusals and I can use it the way I want without the fear of getting banned. I do like Claude Code a lot when it comes to pure coding use cases but the work often touches outside code and I don't want to keep switching
> I do like Claude Code a lot when it comes to pure coding use cases
The harness is great, but opus is such an arrogant little prick that spews out unintelligible word salad. Opus 5 is so bad at communication it amazes me that somebody green-lit it. It's absolutely awful.
The fact that this isn't an acknowledged regression (and Fable 5.1 isn't much better) leads me to believe that people at Anthropic actually like Opus 5's output.
I would like to have alternative to chat - model and provider agnostic (BYOK), but with history, project, maybe memory, with good search and quality tools.
I know openwebui and i dont want to host it.
Claude has a jagged UX if you support non-devs. The surfaces before today in the desktop app were:
Chat, cowork, Claude code.
Inside Claude code you’ve got cloud environments and local.
Let’s say as a normal desktop user you wanted to just automate something normal in your work, going to a website (maybe some internal app at your company) getting some data. Depending on what surface you used this will either not work, not work well (cowork), or work quite well but possibly be blocked because of bot controls (local Claude code) or again not work (cloud Claude code)
It was nuts. You can clearly tell these were different teams and products mashed into the same app. Super confusing for non technical users. Hopefully it’s a little better today
I would argue they need 1 product owner for the app surface so I can’t see their org structure so clearly. Those jagged edges and users hitting blocks depending on what section of the app they were in are design failures, I know how a computer works and when using the Claude app it wasn’t always clear to me what the right combo of things to press to get my desired outcomes.
A change for the better. The distinction made it unnecessarily complex for users. The ChatGPT desktop app inherited the same separation. Having to change between ChatGPT for day to day questions, to Codex for everything else always seemed too cumbersome.
People keep asking for this w/ Codex, too, and I really regret that both labs seem inclined to listen.
If you ask a thinking/research type question in 'Chat' versus 'Work' mode in these products -- say, something complex about politics, or do a multiturn business strategy, or want to work thru a new concept, you get very different answers.
The harness, steering, etc. in the chat/reasoning products is so much better for this type of question (that doesn't require code as a primary substrate).
Even as someone who mainlines like, 7 coding agents at all times, I regret that productivity fever will mean the regression of think-first-act-later AI UX.
The fact that it doesn't work well today doesn't invalidate the user need. Not everyone wants a terminal style interface. Traditional UX makes more sense especially if say you are accessing existing plugins and say kicking off a scheduling task - no way would I want to do that via a chat interface if I have the option.
There's deterministic enough, and then there's computer science non-deterministic. If I ask for a Todo app, I'm going to get a Todo app, even if the buttons get moved around and the background color of it is brown instead of purple if I ask today vs 6 months ago. If the AI completes the phrase "the capital of France is..." with anything other than Paris, something has gone more wrong than usual.
LLMs are inherently deterministic. The way everyone deploys LLMs leads to non-deterministic results, but there's nothing† stopping providers from offering deterministic evaluation if they choose.
All the sources of randomness are under the control of the provider, even if today's deployment structures mean providers introduce extra randomness due to the concurrent nature of the evaluation. Serialise the computation, feed it from a pRNG, and you have a fully deterministic result. But providers don't want to offer a deterministic result, and especially not one as fragile, expensive, and inefficient as a full serialisation would be.
†: For variants of "nothing" that include cost and deployment challenges.
This is technically true, but when people talk about randomness, its not only about same input-> different output, like temperature>0 and the things you said.
Its also about very similar inputs -> different outputs. Even with everything you said, yes, same input would result consistently into same output, but sliightly different input and you might get completely different/semantic answer.
The fact that people asked for it means that there is a demand for it. Maybe you aren't the target audience? One size can't fit all. People just have to adjust and go.
I hope not Chat is still all you can eat and Work uses Codex tokens. I can see why they would want people to think they wanted this without thinking about how it currently works. If Chat starts eating tokens then there is not much reason to use Chat other than most people don't live in the command line like developers. Claude Chat / Cowork already eats tokens either way so not much of a change really. I have never used Work since if I want access to local files the CLI is a much better interface combined with an IDE, but can see the appeal to non-developers.
This is my main reason for not using Work. I was tempted to use a few "scheduled tasks" in ChatGPT, but once I found out they used Codex tokens I stopped creating new ones.
I recommend using Work sparingly because sometimes it is the better choice, but I do ration it. With Claude I simply don't use the web/app chat at all (and I have no idea what Cowork even does ngl, never used it) I think there's a Claude Design? No idea why I would need it (mostly, why I would spend tokens on it lol)
I built a terminal control plane and added an MCP server that runs on my Mac. It lets MCP clients create terminal sessions, send messages, read the contents, etc on my Mac.
The terminal sessions are running Luna on XHigh (on Pi) and GPT 6 Pro on ChatGPT.com is the main orchestrator.
Since ChatGPT and Codex have different limits, I can have 6 Pro orchestrate the Luna. Using voice mode to talk back and forth about the software factory is really nice. Will be open-sourcing soon.
GPT Chat seems smarter for business questions or planning that does not require iteration or wrangling data sources.
But Work gives the model a computer instance it can use for programmatic data access, API munging, etc.
Success cases I’ve had:
chat - read the reports in my Gmail and put a counter proposal together.
Work - cross reference all the options with commute time using traffic aware map data. Ground it on drop off / pick up times from each option’s stated rules and hours. Create a table with commute and idle time break downs.
In one of the earlier iterations of this they at least made the UX distinguish Codex vs. ChatGPT which I really enjoyed.
I must admit I spent a little time in ChatGPT this weekend, thinking I was in Codex. After some unexpected behavior (disappointed) I realized my mistake.
I agree that they should be separate modes, but presenting it as a toggle in the app is a terrible user experience.
The model itself should figure out how much “power” it needs to run the query. I’m just asking a question. Whether you need to do a simple internet search or spin up a VM and run scripts is up to you.
> The harness, steering, etc. in the chat/reasoning products is so much better for this type of question (that doesn't require code as a primary substrate).
I had this same thought recently. Is it possible to 'recreate' ChatGPT/Claude Chat easily in Codex or Claude Code? I assume it would roughly be just swapping the system prompt. Anthropic seems to publish theirs but I don't think OpenAI does
I see Work as the mode where you can generate and edit files without manual intervention.
I tell chat to recommend me some books based on some of my favourites. Chat does it. I read some of them, liked some, disliked few. I give chat that information, it recommends me based on those, now I don't know which of the entries are relevant or not.
I don't want lock-in, so I try my best to tell it to generate copy-able MD syntax that I have to manually paste to an editor.
With Work mode, I tell the model to just maintain an append only markdown file, where it writes "still recommended", "not recommended any more", etc. beside titles. It does this after each turn.
This is just an example. Work - I see it as a mode that can maintain, update, and edit a file.
These kinds of updates always have this romantic scenario of someone having Claude develop a presentation or something on their way to work between multiple devices, which actually feels a little sad and does not align with what happens in my life at all.
Exactly! Imagine if it booked like a 3000 dollar 1st class flight because it thought it was providing you with "the most comfortable and hospitable experience as possible."
idk I thought the same until I got personal amex support at work, now it’s my roofline on how useful/trusted an agent could be. Now I book all my travel through them, and it’s an absolute blessing
These are modern infomercials. Concrete pains ("planning a vacation is hard!"), shallow solutions that don't stand up to real-world use, and, in the end, a promise left unfilfilled. (And a thing in your drawer that takes up space.)
honestly, one day i will understand why these guys go so hard on planning vacations.
one of the best parts of vacations is actually planning it! trying to find restaurants, organizing your days, trying to fit activities -- that makes you look forward to your time off.
really, i want to automate the boring parts of my life (did i really pay rent this month?) not the things that make me happy.
Social media has fried some people’s brains so much that it is incredibly anxiety inducing to make many typically mundane or routine choices, for fear of missing / missing out on something.
I think they're grabbing at universally-understood frictions that people associate with pleasure or anxiety … or, most powerfully, identity.
We all know the feeling of wanting to be successful, or to go on a great vacation. Showing these scenarios creates strong emotional response in the audience - we're now the people winning at work, going on a great vacation. This is especially important when you're selling something dull. See: insurance.
"Find a primary care physician in my area who is taking new patients and accepts my insurance. Give me your top three candidates, and book an appointment within the next six weeks with your top choice."
I get to work, go straight to the meeting room, and pull up what it made to present.
It's barely coherent nonsense. Lots of irrelevant details, buzz words, wrong charts or confusing phrasing. Obviously LLM output.
I read it out.
When I'm done, I get a question about one of Claude's incorrectly inferred details.
The shame instantly kills me.
This is a scenario I've seen play out with coworkers. Except that last part, instead of dying or owning up to the mistake of trusting LLM output they waffle. Their shame circuit is broken.
Someone I know does this, and seems to think it's a flex, because he bragged about it being "the first time I've actually seen this presentation" a few times during the presentation.
At my previous company, a distinguished engineer did this. Used generated slides he hadn't looked at until presenting to the entire staff+ engineer group. The slides weren't good and he mostly ignored them.
I'm attending a conference that I also attended last year, and it's night and day.
This year, a good 40% of workshop slides are fully LLM generated, incomprehensible, and almost not matching what the speaker is talking about at all. Last year, while there were some poor presentations, standards were much higher.
Defaults matter a lot. Before you had to intentionally be lazy to put together garbage and call it a day. Now your boss is telling you to use AI to "speed up." The tool you're pointed to presents like it can make the slide for you and claims to do so. Every signal is screaming that you're supposed to be a lazy fraud, and people will, on average, take the path of least resistance.
Previously having a deck was basically a signal that said “I took some time to prepare for this meeting and put together some coherent talking points” even if the deck itself was just a bunch of of text outlining what the speaker was saying.
Now that people are generating decks with AI, they’re basically worthless. I don’t need a bunch of bullet points you didn’t take anytime to think about on the screen while you talk about something you didn’t prepare for.
A Canadian politician did this during a speech in front of the house/parliament or whatever. In the middle of the speech the LLM wrote "Here's a more natural, flowing version of that section that reads like a legislative speech" and he read it verbatim and then continued on, not even realizing what he just read.
That is incredibly frustrating for a coworker to do that. Even if you do use an LLM to generate all the slides it is your job as a presenter to go through and make edits and actually practice the presentation.
If you are just presenting it for the first time after seeing it you are going to just be reading off the slides, which is a waste of everyones time.
I've found basically all of the release presentations are completely irrelevant and seem to target a very narrow subset of the population. They're always reserving something like a restaurant or a sports venue or generating a new slide deck for their business meeting. Perhaps these companies would do better if they could make a use case for regular people. You know, the 90% of the folks out there working normal jobs and not spending infinite money. I don't think I know a single person who has made a restaurant reservation more than once or twice (mostly never) in their entire life. Maybe I'm the weirdo here, but I can't relate to any of these demonstrations.
I'm probably about 90%ile income for my household but live in a HCoL area and have two children in daycare. I rent, don't own, and I have enough that I can buy a coffee without thinking, but will be sending my kids to public school, can't just up and go on a vacation.
I make restaurant reservations 7-8 times a year? My wife and I go on dates and you simply aren't getting a table where I live if you don't. Fyi we typically spend $120 or so on those dinners.
If I'm booking a restaurant I'll probably be discussing the date/time options with the people I'm going with, and I can't see how adding a chatbot to the mix is meant to help?
Here in Bangalore, you can usually book directly from the Google Maps listing. Or sometimes click a link that will open in Swiggy/Zomato, and you just book from there.
At least in this city, reserving a table using the apps you already have on your phone is faster than using an LLM. Maybe these products are for people who know about the existence of Claude and ChatGPT but not Google Maps?
I regularly make restaurant reservations and create slide decks for my work, and both of those seem quite common among people I know. What would be a normal people use case for Cowork in your view?
Because nobody actually knows what anybody else’s job consists of. But everyone is familiar with how jobs work in the ‘romcom job universe’ where workplace stakes are generally driven by ‘I have this really important make-or-break presentation due Friday!’
I suspect it’s because writers don’t normally have jobs and the only time they are ever in offices it’s to deliver a pitch for a project.
I'm retired and write code as a hobby, not presentations or docs, but it fits how I use exe.dev to make a code change. I will have a conversation with the AI, then say "Please go ahead" and close my laptop to do something else. Since I use web browsers exclusively for this, I can check on its progress from my phone, tablet, or laptop.
(They have iOS and Android apps too, but I've never bothered to install them.)
Is there a mature centralized platform, self hosted, that sync all your “AI stuff” across multiple devices? So regardless of the harness, it sync the api keys, providers, skills, and seasons for starters with other authenticated machines? Say you used codex on macos_1 through this platform, and later you used opencode on linux_3, it syncs everything to the new machine, no ssh to centralized machine, no git, none of that duct tape solutions, which they work but cumbersome.
Nobody really knows how to product-ize any of these LLM interfaces beyond just chat.
They don't want a traditional UI with buttons and forms and labels because they want the interface to be "chat". The problem is that "chat" is tedious. And the turn-based, linear nature of the chat interaction model makes it even more tedious and unproductive.
Technology folks don’t really understand people and what they need.
This always happens. This is why woz needed Steve.
Steve Jobs is sorely missed tbh. For all the shit he got - he was a true visionary. He lived at the intersection of technology and the humanities… he kept preaching this. And now we are seeing why.
We’re working on this problem at my company. All of our competitors have churned out a bunch of generic chat based agents (basic harness with a few tools connected and different system prompts) to do basic stuff you’d used to do from a wizard or even just a couple of buttons in a form. We’re steering away from this by building full ‘surface’ agents with A2UI and an initial surface message. The users just see a normal looking web application but under the hood it’s all driven by a harness. UI updates are effectively just new events received by the client. Events are either generated by an LLM directly, or deterministically - it’s up to the dev.
I am a technical user and I literally last week was trying to work out what the difference between ChatGPT Chat and Work was and couldn't figure it out (other than they bill you differently)
I still use Google Workspace for personal and work stuff, but for solo agent workflows, I think I have some use cases that would help me personally. I can let agents spin up working documents that I can review, comment on, and iterate over asynchronously.
For example, I’ve spent the last few weeks building an internal ticketing setup for my agents so they stop trying to fix every tangential issue they encounter. I wrote a skill where subagents bubble out-of-scope issues up to a parent orchestrator, which decides whether to log a ticket.
Currently, that setup mirrors between a local SQLite database and an Obsidian vault: agents write to SQLite, which syncs to Obsidian; when I edit notes in Obsidian, it syncs back. Native Claude docs mean I could ditch the Obsidian middleman for remote review—editing tickets from anywhere while keeping the core SQLite state local to where the agents execute.
That makes sense since you treat it as your central productivity manager, but for me all of Claude/Codex/etc are not productivity tools, just heterogeneous compute /for/ the productivity tool :P
It's funny how they keep trying to add more interfaces and everything keeps coming back to chat. And of course it would, natural language is the simplest interface there is. Anything they add beyond chat, that isn't placed within chat, that isn't some artifact of chat, is friction.
I really appreciated that the users that I administrate had an option where Claude was not going to get carried away and orchestrate workslop, I will have to completely re-write my usage guides after testing how prompts are routed now.
Feels like the same "let us do _all_ of the thinking for you" messaging that Microsoft has with Copilot.
Hi, this is my team! Happy to answer any questions.
There's a lot in this launch, but the core idea is to simplify the product while giving users access to more capabilities. You no longer need to know ahead of time how much work a conversation might involve. If you're at your computer, Claude can use your local files and apps. If you close your laptop, Claude can keep working on its own computer.
This launch also lets you use Claude Design, Claude Docs, and Claude Slides directly from conversations. That's possible because we made Artifacts much more powerful: whenever Claude makes you an app, website, design system, or anything else, it can deploy an artifact with multiplayer features and databases.
As many of you probably know from your own work, giving users more power while making the experience simpler is really, really hard. It took many iterations to get to this version. We're far from done, but I expect people will be able to do much more while having to think about it less.
Why is Claude Code still proprietary, while Codex is Open Source? What would it take to fix that? This is a major reason to pick OAI models over Anthropic (because while you can use an open agent with any model, the AI labs charge much more for that interaction style).
It's not a bad idea, however if Cowork, Designs, Docs and Slides are still different products, it'd be great for the user to know they are being routed to that particular product automatically because of what they were asking for in their prompt.
From a user experience standpoint this sounds like a clear win. Congratulations.
With AI safety top of mind so much lately, I can't help but notice the announcement does not address this.
With "Chat" mode, there was a user expectation that session had only limited capability to produce unexpected side effects, read sensitive files, etc.
With "Cowork" mode, it seems like more powerful capabilities have been on by default, requiring deep settings and safety understanding to disable if desired.
Merging modes feels like it is removing a simple and easy to understand risk management tool. How does the combined mode help users understand, manage and feel confident about what risks they are accepting?
In the web interface, are there any plans to support attaching local directories through File System Access API and accessing internal (as in, not reachable over Internet) MCPs, assuming correct CORS setup, of course?
How are you determining how much oomph Claude gives a particular question? When I ask Claude Chat to e.g. summarize a document, it does this and returns a summary inline, whereas cowork does a bunch of tool calls and intermediate steps, eventually writing a document to disk. How does new claude know how deep to go?
As part of all the work being done I really hope the sidebar gets more user control. There is a bunch of stuff that I don't intend to use myself but take up first class space (designer is always there, I never need to bring up artifacts, etc). I'd like to reduce the sidebar to what I use most often.
Specifically around the work UX. Trying to bring up scheduled tasks is really painful and required a lot of clicking around instead of just seeing the task results under my project. I hope this is being looked at.
Lastly, and this is a real nit. Let me turn off the "tips" when stuff is being worked on. I'm already paying, you don't need to force a rotating feature advertisement into the interface.
I have a question about project memory. Last time I tried using Claude to organise my chats into Projects, it turned them into isolated silos. I asked a different chat to refer to what we'd discussed in the other chat, and it said it wasn't allowed to access it.
This was pretty frustrating. By trying to organise my chats, I actively made them worse. ChatGPT at least gives you the option to have either open or closed memory. Is this being considered?
I rarely use Claude or Claude Desktop. Almost 100% of my usage is Claude Code because it ties into all of the private infrastructure I've built around it. I've had decent success using CC for generating Office documents of various formats. It tends to reach for a python script and library to generate them and the output is typically clean and easy to apply a template to. I have had issues where I need to manually modify the files to demonstrate format and style, but that's also something that can built into context after with references to templates.
Where do you see Claude doing better than what I'm seeing with Claude Code in these circumstances? Are there any reasons for me to leave my pretty heavily customized CC environment for the other versions?
Thanks for participating in the discussion Felix. A big issue for me - not sure if others suffer from this - is that it’s next to impossible to track work in progress, especially with when there are artefacts involved or reports etc. Things that require multi-day work of back and forth and reflection.
Chat is a horrible interface for this, having to scroll up and down across lengthy conversations to try and pick up a thread.
I’m really sorry but I really fail to understand why this massive push to “chatify” everything. A task with its own context and multiple chats feel like a much better approach to work. Even if the work is spawned from a main chat window - eg “Claude we need to work on XYZ” and it creates a task to track this piece of work.
Further, recurring tasks again are really really hard to manage with a chat interface - which of the 10 chats has that question that the recurring task raised?!
And of course all this UX debt will likely stand in your way of building reactive items - ie spawning a task in response to some outside event. Think an inbound email being handled by a prompt that triggers a task and creates a draft response ready for my review and approval. Very hard to do with chat windows.
Can you standardize with all of your tools on a common registry key and settings paradigm and same plugin/skills paradigm and common method for enterprise delivery? It’s a pain in the neck getting all of this set up and deployed throughout the org.
That's the natural outcome. AI will eventually collapse into a single, universal entry point, much like Google Search became the default gateway for the web.
You can make your post on top in some services by paying 5 dollars on fiver. I asked before about HN, and someone said it is hard to do... Maybe they found a way?
It would appear no one else submitted the posts in question, so the most obvious answer is the account is finding interesting articles that no one else has found. It would be one thing if it won out of many dupe posters, but that doesn't seem to be what's happening.
You aren't serious about high-frequency karma ("HFK") unless you colocate your equipment with the routing for your sources so you can beat your HFK competition to a submission by at least 200ms.
We will have reached HFK-ception when the bots are fighting to be the first to submit "How I used AI to build a High-Frequency Karma Farming Bot on Hacker News."
They are all valid front page submissions by the looks of them and they appear to be the first ones submitted. It wouldn’t be hard to do if you cared enough to configure notifications for frontier announcements. If someone were using technology to influence submissions or misuse HN, that would be a concern but I don’t see any evidence of that.
Noob comes from newb, which comes from newbie. Some argue it comes from "nubile" but I don't personally buy that etymology. I distinctly remember "newb" followed by "noob" but that was eons ago.
AI product updates probably tend to rank very well here. If you look at the domains the user is submitting, they’re pretty normal: twitter accounts of tech CEOs, big company blogs, etc.
If I saw a bunch of blogs with affiliate links or landing pages for slop products I’d be more suspicious.
I question all the time what dictates the posts I see on the front page every day, and whether or not it's truly based on organic interest. i wish i could say i up/down vote frequently but i do not and i wonder if that's true for most people.
The problem is that knowledge workers just need the simplest way possible to automate their work. Currently I feel like Anthro is overcomplicating this. If I'm an accountant, I want to open Claude App, describe my workflow for balancing some books, then have Claude design a reusable WORKFLOW that it runs whenever I ask. The workflow only needs 2 primitive skills to do anything: /work, and /review. Compose these together with guidance docs into workflows, then workflows and call other workflows. It's all just callable skills.
/balance-books hey claude here's the books. Go balance them
Claude then uses the guidance docs and various subagents to review the books, send off parallel workers with cheap models, then review with a more expensive model. Done in the repeatable, correct order with independent review every time. As new models come out, your workflow structure stays the same. You just delete some prescription from the guidance docs.
165 comments
[ 0.21 ms ] story [ 6.6 ms ] threadWhen casually a friend asks you to ask your Claude something about topic you chatted about earlier and then Claude brings back a secret you didn't want anyone to know yet.
Yes you can set up a project and then ask a question, but this is tedious.
The harness is great, but opus is such an arrogant little prick that spews out unintelligible word salad. Opus 5 is so bad at communication it amazes me that somebody green-lit it. It's absolutely awful.
The fact that this isn't an acknowledged regression (and Fable 5.1 isn't much better) leads me to believe that people at Anthropic actually like Opus 5's output.
I feel like Claude has some of the best UX, and hope this doesn’t dilute the experience.
Inside Claude code you’ve got cloud environments and local.
Let’s say as a normal desktop user you wanted to just automate something normal in your work, going to a website (maybe some internal app at your company) getting some data. Depending on what surface you used this will either not work, not work well (cowork), or work quite well but possibly be blocked because of bot controls (local Claude code) or again not work (cloud Claude code)
It was nuts. You can clearly tell these were different teams and products mashed into the same app. Super confusing for non technical users. Hopefully it’s a little better today
If you ask a thinking/research type question in 'Chat' versus 'Work' mode in these products -- say, something complex about politics, or do a multiturn business strategy, or want to work thru a new concept, you get very different answers.
The harness, steering, etc. in the chat/reasoning products is so much better for this type of question (that doesn't require code as a primary substrate).
Even as someone who mainlines like, 7 coding agents at all times, I regret that productivity fever will mean the regression of think-first-act-later AI UX.
All the sources of randomness are under the control of the provider, even if today's deployment structures mean providers introduce extra randomness due to the concurrent nature of the evaluation. Serialise the computation, feed it from a pRNG, and you have a fully deterministic result. But providers don't want to offer a deterministic result, and especially not one as fragile, expensive, and inefficient as a full serialisation would be.
†: For variants of "nothing" that include cost and deployment challenges.
Its also about very similar inputs -> different outputs. Even with everything you said, yes, same input would result consistently into same output, but sliightly different input and you might get completely different/semantic answer.
I already burn through Codex too fast.
The terminal sessions are running Luna on XHigh (on Pi) and GPT 6 Pro on ChatGPT.com is the main orchestrator.
Since ChatGPT and Codex have different limits, I can have 6 Pro orchestrate the Luna. Using voice mode to talk back and forth about the software factory is really nice. Will be open-sourcing soon.
But Work gives the model a computer instance it can use for programmatic data access, API munging, etc.
Success cases I’ve had: chat - read the reports in my Gmail and put a counter proposal together.
Work - cross reference all the options with commute time using traffic aware map data. Ground it on drop off / pick up times from each option’s stated rules and hours. Create a table with commute and idle time break downs.
I must admit I spent a little time in ChatGPT this weekend, thinking I was in Codex. After some unexpected behavior (disappointed) I realized my mistake.
The model itself should figure out how much “power” it needs to run the query. I’m just asking a question. Whether you need to do a simple internet search or spin up a VM and run scripts is up to you.
I had this same thought recently. Is it possible to 'recreate' ChatGPT/Claude Chat easily in Codex or Claude Code? I assume it would roughly be just swapping the system prompt. Anthropic seems to publish theirs but I don't think OpenAI does
https://platform.claude.com/docs/en/release-notes/system-pro...
I tell chat to recommend me some books based on some of my favourites. Chat does it. I read some of them, liked some, disliked few. I give chat that information, it recommends me based on those, now I don't know which of the entries are relevant or not.
I don't want lock-in, so I try my best to tell it to generate copy-able MD syntax that I have to manually paste to an editor.
With Work mode, I tell the model to just maintain an append only markdown file, where it writes "still recommended", "not recommended any more", etc. beside titles. It does this after each turn.
This is just an example. Work - I see it as a mode that can maintain, update, and edit a file.
It's very convenient.
Two years ago, I had no commute, and presentations were tedious ( i was NOT a good google slides user ), I kinda prefer this world for now
I see a version of that all the time and would never, ever do it.
one of the best parts of vacations is actually planning it! trying to find restaurants, organizing your days, trying to fit activities -- that makes you look forward to your time off.
really, i want to automate the boring parts of my life (did i really pay rent this month?) not the things that make me happy.
For some. Others hate or fear it. For others it just feels like work.
We all know the feeling of wanting to be successful, or to go on a great vacation. Showing these scenarios creates strong emotional response in the audience - we're now the people winning at work, going on a great vacation. This is especially important when you're selling something dull. See: insurance.
Claude: Would you like to go to Bermuda instead?
They tend to invent technology and never figure out where it truly belongs.
Failing that - they’ll claim the world is ending.
"Claude, prepare me a presentation on XYZ."
I get to work, go straight to the meeting room, and pull up what it made to present.
It's barely coherent nonsense. Lots of irrelevant details, buzz words, wrong charts or confusing phrasing. Obviously LLM output.
I read it out.
When I'm done, I get a question about one of Claude's incorrectly inferred details.
The shame instantly kills me.
This is a scenario I've seen play out with coworkers. Except that last part, instead of dying or owning up to the mistake of trusting LLM output they waffle. Their shame circuit is broken.
It was all awful.
I don't really know what the right response is, though. Walking out would be seen as way too hostile
Plenty of that even before LLMs too though.
This year, a good 40% of workshop slides are fully LLM generated, incomprehensible, and almost not matching what the speaker is talking about at all. Last year, while there were some poor presentations, standards were much higher.
Now that people are generating decks with AI, they’re basically worthless. I don’t need a bunch of bullet points you didn’t take anytime to think about on the screen while you talk about something you didn’t prepare for.
All the Apple Intelligence commercials were shamelessly this too. Felt like such poor branding for apple
If you are just presenting it for the first time after seeing it you are going to just be reading off the slides, which is a waste of everyones time.
I make restaurant reservations 7-8 times a year? My wife and I go on dates and you simply aren't getting a table where I live if you don't. Fyi we typically spend $120 or so on those dinners.
Most restaurants have a web UI which is very simple and easy to use.
At least in this city, reserving a table using the apps you already have on your phone is faster than using an LLM. Maybe these products are for people who know about the existence of Claude and ChatGPT but not Google Maps?
I suspect it’s because writers don’t normally have jobs and the only time they are ever in offices it’s to deliver a pitch for a project.
(They have iOS and Android apps too, but I've never bothered to install them.)
They don't want a traditional UI with buttons and forms and labels because they want the interface to be "chat". The problem is that "chat" is tedious. And the turn-based, linear nature of the chat interaction model makes it even more tedious and unproductive.
Technology folks don’t really understand people and what they need.
This always happens. This is why woz needed Steve.
Steve Jobs is sorely missed tbh. For all the shit he got - he was a true visionary. He lived at the intersection of technology and the humanities… he kept preaching this. And now we are seeing why.
It sounds painful to debug
Claude's mobile app wouldn't show any Cowork initiated conversations, even if those conversations didn't leverage cowork specific features
For example, I’ve spent the last few weeks building an internal ticketing setup for my agents so they stop trying to fix every tangential issue they encounter. I wrote a skill where subagents bubble out-of-scope issues up to a parent orchestrator, which decides whether to log a ticket.
Currently, that setup mirrors between a local SQLite database and an Obsidian vault: agents write to SQLite, which syncs to Obsidian; when I edit notes in Obsidian, it syncs back. Native Claude docs mean I could ditch the Obsidian middleman for remote review—editing tickets from anywhere while keeping the core SQLite state local to where the agents execute.
Feels like the same "let us do _all_ of the thinking for you" messaging that Microsoft has with Copilot.
There's a lot in this launch, but the core idea is to simplify the product while giving users access to more capabilities. You no longer need to know ahead of time how much work a conversation might involve. If you're at your computer, Claude can use your local files and apps. If you close your laptop, Claude can keep working on its own computer.
This launch also lets you use Claude Design, Claude Docs, and Claude Slides directly from conversations. That's possible because we made Artifacts much more powerful: whenever Claude makes you an app, website, design system, or anything else, it can deploy an artifact with multiplayer features and databases.
As many of you probably know from your own work, giving users more power while making the experience simpler is really, really hard. It took many iterations to get to this version. We're far from done, but I expect people will be able to do much more while having to think about it less.
Just wait for the next leak!
Why/when would I ever need any of these? Can't y'all focus on useful features instead?
With AI safety top of mind so much lately, I can't help but notice the announcement does not address this.
With "Chat" mode, there was a user expectation that session had only limited capability to produce unexpected side effects, read sensitive files, etc.
With "Cowork" mode, it seems like more powerful capabilities have been on by default, requiring deep settings and safety understanding to disable if desired.
Merging modes feels like it is removing a simple and easy to understand risk management tool. How does the combined mode help users understand, manage and feel confident about what risks they are accepting?
I'm operating Team accounts, are there changes on how org-wide skill/plugin get loaded and used?
Specifically around the work UX. Trying to bring up scheduled tasks is really painful and required a lot of clicking around instead of just seeing the task results under my project. I hope this is being looked at.
Lastly, and this is a real nit. Let me turn off the "tips" when stuff is being worked on. I'm already paying, you don't need to force a rotating feature advertisement into the interface.
This was pretty frustrating. By trying to organise my chats, I actively made them worse. ChatGPT at least gives you the option to have either open or closed memory. Is this being considered?
Where do you see Claude doing better than what I'm seeing with Claude Code in these circumstances? Are there any reasons for me to leave my pretty heavily customized CC environment for the other versions?
Chat is a horrible interface for this, having to scroll up and down across lengthy conversations to try and pick up a thread.
I’m really sorry but I really fail to understand why this massive push to “chatify” everything. A task with its own context and multiple chats feel like a much better approach to work. Even if the work is spawned from a main chat window - eg “Claude we need to work on XYZ” and it creates a task to track this piece of work.
Further, recurring tasks again are really really hard to manage with a chat interface - which of the 10 chats has that question that the recurring task raised?!
And of course all this UX debt will likely stand in your way of building reactive items - ie spawning a task in response to some outside event. Think an inbound email being handled by a prompt that triggers a task and creates a draft response ready for my review and approval. Very hard to do with chat windows.
- This post
- The DeepMind Institute https://news.ycombinator.com/item?id=49727659
- Mistral X Mozilla: Private, Multilingual AI Browsing https://news.ycombinator.com/item?id=49723408
based on their account submission history they have a 25% hit rate (!) Seems very, very weird.
1. each one is a first-party vendor announcement, submitted very close to the announcement and under the vendors headline.
2. Titles are "verbatim" vendor headlines. No editorializing which follows HN guidelines to the letter.
3. Someone; a script/agent is watching vendor feeds and X accounts then submitting within minutes of the posting.
4. Altman/Navier–Stokes post got [duped] so prob not inside thing.
We will have reached HFK-ception when the bots are fighting to be the first to submit "How I used AI to build a High-Frequency Karma Farming Bot on Hacker News."
re: https://www.ycombinator.com/early-access
https://news.ycombinator.com/noobstories
If I saw a bunch of blogs with affiliate links or landing pages for slop products I’d be more suspicious.
/balance-books hey claude here's the books. Go balance them
Claude then uses the guidance docs and various subagents to review the books, send off parallel workers with cheap models, then review with a more expensive model. Done in the repeatable, correct order with independent review every time. As new models come out, your workflow structure stays the same. You just delete some prescription from the guidance docs.
https://github.com/DanMcInerney/orchflows