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.
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)
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.
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.
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.
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.
Yes the point is that they wouldn't have to adjust unless they are merged into one. Precisely because one size can't fit all, yet they insit on one size now.
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.
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'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.
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.)
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
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.
There are a lot of people using this outside of the coding space, most of my friends use these to create decks and things like that to present. Mind you they never double check information and numbers.
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.
Will there be a Claude Sheets too? I find myself using Claude to work with spreadsheets more than almost any other document type. It does okay now, but feel like it could be even better, especially with visualizations and formulas.
I liked the ability to explicitly only chat, without the possibility of Cowork activating since I have never wanted that. It's unclear how I can still guarantee that now.
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.
I don't think it's that anyone wants the interface to be chat. I think it's that the underlying technology is inherently words in and words out. It's similar to how devices with capacitive screens are most naturally going to support tapping, dragging, and pinching.
It's funny to me that instead of moving existing tech to a word in word out paradigm, people are dropping millions getting the chat to work with the existing paradigm.
Probably inevitable, but seems like a lot of disruption could happen there.
My personal hot take is that the product people (engineers too probably) at these companies are just straight up lazy. Yes, a new UI paradigm is hard, but it's been painfully obvious that chat just absolutely sucks. It's also obvious that some DSL-ish thing is possible, something that just does token juggling and the end-user sees some UI behavior.
I know this all sounds abstract. I've been mulling over it for the past year and it's very hard; and LLMs are super janky and inconsistent so it's 100% not trivial. So in some sense I understand why a lazy bottom-of-the-barrel "chat interface" has become the de facto standard.
The great confusion for Anthropic and OAI is how to break out from pure chat without signalling you in fact intend to eat all of your customer's business too.
IMO no one knows what LLMs are capable of yet and the goal post keeping moving every few months that building any specific UIs right now risks rendering them obsolete or too slow.
e.g. We went from somewhat smarter code autocomplete, to asking chatgpt copy paste, to cli agent running inside your project, managing session, to GUI to manage that, to projects where you talk to a "Chief of Staff" agent that manages other sessions, to who knows what's next.
I think the right interfaces for LLMs right now need to be very simple and easy to change/evolve. And chat still seems to be the best default solution.
> The moment someone figures out a new modality for LLMs is when we’ll see the next hockey stick.
Honestly, that fills me with fear. LLMs exist to make money to their companies, and said companies are not gonna turn around and say, "you know what, go are going to make an android for each elderly person, that can not only help them with their medications, but that can actually make their medications, tailored to their biologies." Instead, they are going to go for the low-handing fruit of "you know Bob, the guy who makes jokes in meetings but who is grumpy about delivery timelines? Well, we are going to make an android to replace Bob. MetalBob will make even better jokes. The blue model will be able to explain in excruciating level of detail why timelines aren't reasonable. The red model will walk through the cubicles with a whip to ensure everybody keeps working all the time, and nobody goes to pee."
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.
One thing I would like is a tree-like chat structure like reddit / HN. Many times I abandon the direction things have gone but would like to resume at some ancestor or sibling response.
>Each conversation is a Yjs CRDT tree. It can branch into sub-threads (recursively), and you can drill down, backtrack, edit, undo/redo, and inspect everything: tool calls, approvals, and the raw context JSON going to the model, etc.
I solved this with task management and got work trees.
The downside of course is branching in got, but I usually don’t go of course more then a handful of tasks.
I had to engineer my own ticket management to keep opus on target. It’s been great for managing work, history, audit trails and commits are tagged with the task id.
I implemented subtasks to deal with the way Claude likes to stage its own objectives.
Then I made enforcement logic in the task manager so tasks can’t be closed out until reviewers have consensus on the same sha.
The adhd that is Anthropic demanded I build it and now I’m knocking out issues faster than Batman.
Before OpenAI dropped ChatGPT, nobody knew chat would be such a hit. Several labs had versions of these things, Google engineers were getting finessed by a rudimentary model just like in the movie Ex Machina that came out 10 years prior which was satirizing Google
But I don't agree that they aren't product-ized. There are many applications doing calls to LLMs behind the scenes and are hits, leveraging structured data very heavily and not having conversations with users at all. I would say that there is a predictable scope creep from executives to surface a conversational aspect though. We need to bring representation to that so we can point to some other best practice to push back
API calls are expensive and slow, so I’ve shied away from integrating them into my personal projects. And of course there’s always the risk of hallucination.
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.
I turned off Anthropic properly and switched all of that over to OpenAI yesterday. (I use other models for other things too, especially DeepSeek in Pi).
Honestly aside from the voice it uses you wouldn't notice a difference. Switching costs are low, vote with your wallet.
> I turned off Anthropic properly and switched all of that over to OpenAI yesterday.
Same, just a while longer ago.
I much prefer how OpenAI models write to Anthropic's, will probably revisit Anthropic in a generation or two. Context size is more limited, but no critical forgetfulness due to compaction so far, though I also like having plan files around both for future reference and improving chances of success at long form work.
Also tried out Kimi K3, was nice but slow (and apparently routed some requests to Claude anyways), GLM 5.3 was faster and still pretty good but the token allowances were kinda limited.
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
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.
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.
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
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.
The mistake here is thinking or treating AI tools as special or different in any way. We’ve had tooling that does this for decades. Ansible, puppet, saltstack, NixOS/home-manager, and dozens of other solutions. Even better, writing the configuration is extra simple now, since you can just ask your agent to do it.
True but that’s not the same, think of it like a git but for AI related, I just open whatever harness, I authenticate to that centralized server, and it pulls everything and continue from where I was done last time, seamless and secure. Right now personally I have multiple machines, and I use key managers for the secret API, git for the skills, and sessions? Manually imported/exported. The alternative is using ssh (or herdr or others) and remotely accessing a specific machine where it has everything you need, but it’s not always reliable and better to avoid single point of failure too if anything goes wrong.
There are several projects that injest coding sessions from lots of agents, but for the chat side I haven't found much, except replacements for Claude Desktop. Jan, for example.
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.
My point with above is to criticize their strategy, I have been losing faith on Anthropic and feeling like they are throwing everything at the wall to see what sticks in a more desperate fashion rather than careful one. This sort of decision to me, alongside all their other initiatives recently with shipping agentic PaaS rather than enabling others to build with their tools has been a miss on thoughtful leadership. Sorry for the snark take, thank you for moderating.
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.
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.
I already burn through Codex too fast.
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.
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.
"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.
Plenty of that even before LLMs too though.
"What does this bit mean?" "I dunno."
I see a version of that all the time and would never, ever do it.
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
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/when would I ever need any of these? Can't y'all focus on useful features instead?
Not really proving their usefulness there :P
Just wait for the next leak!
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.
Probably inevitable, but seems like a lot of disruption could happen there.
I know this all sounds abstract. I've been mulling over it for the past year and it's very hard; and LLMs are super janky and inconsistent so it's 100% not trivial. So in some sense I understand why a lazy bottom-of-the-barrel "chat interface" has become the de facto standard.
e.g. We went from somewhat smarter code autocomplete, to asking chatgpt copy paste, to cli agent running inside your project, managing session, to GUI to manage that, to projects where you talk to a "Chief of Staff" agent that manages other sessions, to who knows what's next.
I think the right interfaces for LLMs right now need to be very simple and easy to change/evolve. And chat still seems to be the best default solution.
Traditional and boring works for me.
The moment someone figures out a new modality for LLMs is when we’ll see the next hockey stick.
Honestly, that fills me with fear. LLMs exist to make money to their companies, and said companies are not gonna turn around and say, "you know what, go are going to make an android for each elderly person, that can not only help them with their medications, but that can actually make their medications, tailored to their biologies." Instead, they are going to go for the low-handing fruit of "you know Bob, the guy who makes jokes in meetings but who is grumpy about delivery timelines? Well, we are going to make an android to replace Bob. MetalBob will make even better jokes. The blue model will be able to explain in excruciating level of detail why timelines aren't reasonable. The red model will walk through the cubicles with a whip to ensure everybody keeps working all the time, and nobody goes to pee."
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.
>Each conversation is a Yjs CRDT tree. It can branch into sub-threads (recursively), and you can drill down, backtrack, edit, undo/redo, and inspect everything: tool calls, approvals, and the raw context JSON going to the model, etc.
The downside of course is branching in got, but I usually don’t go of course more then a handful of tasks.
I had to engineer my own ticket management to keep opus on target. It’s been great for managing work, history, audit trails and commits are tagged with the task id.
I implemented subtasks to deal with the way Claude likes to stage its own objectives.
Then I made enforcement logic in the task manager so tasks can’t be closed out until reviewers have consensus on the same sha.
The adhd that is Anthropic demanded I build it and now I’m knocking out issues faster than Batman.
Often times I just sort of wish to undo.
But I don't agree that they aren't product-ized. There are many applications doing calls to LLMs behind the scenes and are hits, leveraging structured data very heavily and not having conversations with users at all. I would say that there is a predictable scope creep from executives to surface a conversational aspect though. We need to bring representation to that so we can point to some other best practice to push back
API calls are expensive and slow, so I’ve shied away from integrating them into my personal projects. And of course there’s always the risk of hallucination.
"Sure! And since you were asking me about lobsters yesterday, I'll make it red and seafood themed!"
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.
Honestly aside from the voice it uses you wouldn't notice a difference. Switching costs are low, vote with your wallet.
Same, just a while longer ago.
I much prefer how OpenAI models write to Anthropic's, will probably revisit Anthropic in a generation or two. Context size is more limited, but no critical forgetfulness due to compaction so far, though I also like having plan files around both for future reference and improving chances of success at long form work.
Also tried out Kimi K3, was nice but slow (and apparently routed some requests to Claude anyways), GLM 5.3 was faster and still pretty good but the token allowances were kinda limited.
[1] https://m.youtube.com/watch?v=BUAnyVAanac&ra=m
You need to use frontier models to understand where the puck is going, but also to use local ones for anything remotely sensitive.
Feels like the same "let us do _all_ of the thinking for you" messaging that Microsoft has with Copilot.
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
Genuinely curious.
There are several projects that injest coding sessions from lots of agents, but for the chat side I haven't found much, except replacements for Claude Desktop. Jan, for example.
Claude's mobile app wouldn't show any Cowork initiated conversations, even if those conversations didn't leverage cowork specific features
"Please don't post shallow dismissals, especially of other people's work. A good critical comment teaches us something."
https://news.ycombinator.com/newsguidelines.html