Wowza this will be a token guzzler. Assuming Claude is parsing every message posted on multiple slack channels, compacting knowledge etc.
Looks like Anthropic is progressing further into platform territory and conquering Agentic use cases left right and centre. If you’re building an agent platform for workforce productivity today, your best best is model agnosticism and focus on token cost control.
I don't understand how this is gonna fly for enterprise security and compliance. Claude needs to inherit permissions from somewhere, and those permissions will never align with the members of a slack channel. And finding the lowest common denominator of access probably results in a dumbed-down, useless experience.
The only way it works is if customers truly start treating agents as humans with the same liability as an employee.
An admin scopes permissions on a per-channel basis. It doesn't allow external actions until an owner specifically provisions that tool for that channel. I think.
But people can be invited to a channel after @Claude is provisioned. So yeah, I suppose you'll need to be deliberate about channel memberships.
Which is a bad pattern. Around here, you can be granted access to most channels just with vague reasons for why you need to be in there. This is a disaster. Culture will degrade. Suspicions will grow. Security theater.
The most important difference from other products:
> @Claude is multiplayer. Within a given Slack channel, there’s one Claude that interacts with everyone. This means that anyone can see what it’s working on, and can pick up the conversation from where the last person left off. This makes tagging Claude very different from working within a single chat or for a single task—it’s much more like interacting collaboratively with a teammate.
We have the opposite problem. We've tried this slack integration (first with NanoClaw, then Hermes, but now just building our own), and we actually want the OPPOSITE.
In private context, I want to have a per-person conversation with durable context for that person's private chats. I also want that person's permissions to extend. Like contractors in our slack should only be able to ask and get back information about clients that they've been attached to, not our entire knowledgeable.
And we've implemented ALL that, but just with a lot of custom code. We've put in interceptors that put in per-user keys into the MCP connection so only certain tools are even exposed, etc.
Hopefully there is some sort of version history implemented or planned like they have in Cowork[0]... this sounds great until a co-worker hijacks your Claude session with a worse idea and derails it from what you were intending.
[0] "Editing this message will create a new conversation branch. You can switch between branches using the arrow navigation buttons."
I don't think that that's quite what the parent commenter had in mind.
From reading that and materials on it, it seems unclear if – let's say you do what's done in the demos on the site and 'dispatch work' from a thread in a shared channel (e.g. from some discussion) – that if any one of your coworkers replies below you and says, "Actually, could you fold in <blah> as well?" that Claude wouldn't listen to them and thus derail the work.
Yeah, there is a level of organizational trust that is required to use this tool (as with any system that allows distributing access via service accounts).
We do signal to Claude that there's a difference between a conversation's initiator versus incoming participants and we've found that in situations where people disagree on an approach, Claude patiently waits for a resolution while correcting any misunderstandings.
It's also worth mentioning that since Claude has its own identity, a coworker cannot enter a thread and commandeer _your_ identity; you collectively steer how Claude acts with its _own_ identity (it opens PRs as itself, browses Datadog as itself, etc).
The killer feature here is an auto assembling todo list, punchlist, project management.
Discussing what needs to be done next, and having it automatically sort it into what subtask it is, if it applies or is blocking to multiple other tasks. Recording specifications, measurements, dimensions. Being able to ask other people facts and have them correctly documented into the right task.
Company Brain / Knowledgebase is imho more rear facing, whereas todo is future facing.
Thats a double-edged sword in some scenarios. If you're trying to keep info private then feeding it into a shared agent basically means that you can't guarantee that privacy. I'd imagine the approach here would be to have separate agents for private data and then restrict Slack access, but I could imagine tons of accidents from managers that habitually @Claude without understanding the implications.
Every channel has its own Claude, and Claude's access is configurable per-channel and per-workspace. Private channels don't leak information to other channels
It's interesting to see the market movements from the big players.
This announcement is all about Claude extending its reach beyond single-player workflows and into multi-player workflows.
On the flip side, Slack just announced MCP support for the Slackbot AI chat capability embedded within Slack. It is, for now, exclusively single-player.
Single-player is the "safe place" (relatively speaking). The context, permissions, and standards (MCP/MCP UI apps) all work reasonably well for it, but get super complex or break down entirely when thrown into a multi-player shared context. I suspect Slack is doing what they are doing with an eye towards multi-player, but it's hard to say how that will manifest.
For a real life example of this challenge: I work in scheduling (for Reclaim.ai) and you can ask our chat to find time to meet with a coworker and we go find time and help explain why certain times won't work. For example, it might say: "I couldn't do 11am tomorrow because you've got a job interview scheduled on your personal calendar". This is safe and fine to do in a private context.
But... imagine if one were to ask our service (or Claude) to find time with someone and it replied to the thread for everyone to see: "The soonest I could find is 12pm tomorrow. Reggie is available at 11am, but Lightbody has a job interview so it won't work". WHOOPS.
I think the other comments in this thread have the right idea of it: for this to really work safely, the permissions model needs to be nailed down, and it may mean that you end up with multiple identities of "Claude Tag" (or whatever agent you engage with in a public forum), and the context it gets is only the context that particular identity is entitled to, just like any other employee. But then that gets tedious because now I've got even MORE "people" to keep track of and know who to engage with, which is half the problem getting work done in large enterprises.
Will be interesting to see how this evolves. I'll have my popcorn out :)
If your mixing work with personal calendars that’s a bigger problem.
Work should be entirely separate calendars with things from personal added to it in generic blocks the owner can understand but protects from this leakage. Work calendars are owned by companies (generally) and should be treated like the are looked at by coworkers.
WRT “Claude learns over time” - this is the biggest gap for me in the current system. As I scale my usage of Claude at work, I observe that it’s quite bad at distinguishing what it should “learn” (memorize) from experimental or just wrong data. It builds and builds on a foundation of sand, making sometimes hidden assumptions and turning them into actionable insight thats just not correct.
It recently wrote an entire dissertation for an epic, assuming it was related to some other project, where it had earlier made the wrong guess about a vendor capability (from their marketing materials, no less), and it all had to be thrown away. I cleared the memory, but it appears to be still pulling from some corporate data source i cant control or locate.
How are Anthropic's production applications poor quality? Other than memory use (which unfortunately is the industry standard nowadays) Claude Code is pretty solid.
Claude code is by far the buggiest piece of software I interact with, If the underlying model weren’t so good, I would never opt to use it.
It takes multiple seconds to launch, random lines disappear in the scroll back, it’s internal state gets messed up causing the TUI to show duplicate and/or offset lines, and it frequently causes some kind of GPU buffer corruption causing the entire terminal env to show garbage.
Yes, this is the ubiquitous memory issue that I mentioned. Unfortunately, it is now the baseline in all modern apps.
>random lines disappear in the scroll back, it’s internal state gets messed up causing the TUI to show duplicate and/or offset lines,
I haven't seen this issue, other than when I am using a shell that is bugged and does it with all TUI/console programs (usually a virtualized shell which I resized). Do you have a reproducible example of this?
Their "safety" system that arbitrarily flags things.
Tons of examples of innocuous strings setting it off and sometimes with financial impact like the OpenClaw/hermes thing (just having the word "hermes" would insta deplete your quota and start charging you API rates in extra usage)
Their product features are a venn diagram, so depending on the tool you use, you may or may not have access to a "Claude" feature
i.e.
- design is only available in web
- cowork is only available in desktop, sharing projects only works in chat, not cowork, which is arguably the more valuable place to have that feature (re: multiplayer like tag). SSO only works if you type your email, the "login with google" button be damned, and only after you finish typing your email does the login button text change.
- cli has a number of features only available there, with the cowork equivalent having a different name iirc
If you admin/support other people using the breadth of tools, you will see more of the slop they sling
It's surviving *for now*. This pace isn't sustainable, they know it, but they need to pretend until the IPO. That's when they'll start worrying about quality. The pace of feature creation is going to decrease the moment they float.
From the little I know from my interactions with their staff, it’s something Anthropic sees as critical. They are well aware their API isn’t reliable and that it is and will cause troubles
Not germane to me whether it holds them back or not, I don't even use their products anymore. I'm just taking the piss, having a cheeky laugh at the expense of the company who brought us such breathless prose as "it’s a bit like selling nuclear weapons to North Korea" and "we are releasing a model that is too powerful for the public".
Like okay Jan lmao, seems like the North Koreans take their uptime a little more seriously than the nuclear weapons developers at Anthropic.
This is quite brilliant. This will seamlessly integrate with every collaboration systems as long as they are able to “enroll” Claude as a user. Claude gains knowledge across systems and makes the ai enhancements for each app/system unnecessary. For example if slack creates something similar it’s less useful because Claude’s model works across any system/app. More so, anthropic can either charge for it separately or not , companies will pay in extra used tokens anyway.
Is someone here using a Claude product that's not code? I'm puzzled about the amount of products they put out. I know a lot of people using Claude but we're all using the terminal-based code. Even for non-engineering stuff it suits great (tax documents, 3D modeling with blender through MCP, academic research, etc.)
And it sometimes feels like these teams are not talking to each other. Using Claude Design? The way to hand it over to Claude Code is to download a .zip file with the html mockup and some description, which you then have to copy into Claude Code so it can use it.
The latest version of Claude code as of yesterday or day before has Claude design Mcp so you don’t need to do this now.
edit: note this may not be official release, and may be unavailable for some users. I saw it show up yesterday listed as available Mcp and used it to view projects in Claude design.
still doesn't feel integrated, like it could have been built by any other company, e.g. Microsoft could have built Microsoft Design and it would look better integrated.
For example, it writes the whole front-end twice, once is claude design and then later it has to read it again and re-implement it in code. Also a lot of stuff (e.g. Claude.md, skill files, etc) are not supported, and they have their own set of ui-design and design systems, which claude code doesn't support.
I think Claude Design is a wonderful product, i'm just pointing that it is an independent product to Claude Code, heck even today it works better with Codex than Claude Code (that is how i use it with my own browser-use agent, i tell it to browse the design in Claude.ai/design and re-implement the design, works much better than downloading the zip file and asking the model to implement in that way.
I use Claude Desktop (& essentially equivalent mobile app) to ask frivolous aspie questions about things society long ago accepted. I enjoy its responses, interpret how you wlll. (my claude.md file has instructions to tell it that the premise of any question i ask is as likely as not wrong, and to never be sycophantic).
But beware the userMemories file (ask Claude to give you a dump of it). bonus points if you can figure out where that file lives.
At first I was freaked out that usermemories is a subpoena’able psychological profile on me. Then I realized that the same file will be produced for spooks all over Langley, so that the day Anthropic gets hacked those profiles will see the light of day (caveat along with the rest of the user base of course) and so I felt more catharsis as a result.
> At first I was freaked out that usermemories is a subpoena’able psychological profile on me. Then I realized that the same file will be produced for spooks all over Langley,
The spooks at Langley aren't going to produce such a file for run-of-the-mill encounters that you have with the legal system, but Anthropic will.
At my work we rolled out Cowork to all the non-technical staff. People are using a ton, wired up for read access to M365, Confluence, etc. as sort of a psuedo enterprise RAG + all the document creation/file management it can do.
I know my users would actually like Claude Tag, but unfortunately we are in Teams, not Slack, as are most other non-tech companies.
Cowork/Claude Desktop itself is also quite a frustrating product. There's no native audit log unless you basically wire up your own with the API & a log aggregator. You can't selectively enable Claude Code access per team member, it's all or nothing. Some of the MCP connections (like QuickBooks Online) don't do RBAC at all, it's all or nothing for every user in the team.
Maybe enterprise isn't their target market, but they do keep making features that make it seem like they want that market. But if they do, they really need to step up work on governance features & RBAC for specific features and settings per member in the team.
If they don't, Microsoft will eat their lunch for enterprise non-programming use.
This happened at Google because there aren’t enough engineers to maintain all these services. This situation may not apply to Anthropic, as they can set up features to be maintained in perpetuity by Claude.
Our entire org migrated from random ChatGPT accounts to all Claude Desktop specifically targetting business users. So a lot of non-dev teams in our org see it as essential.
They are going after the non-developer side of the business. Many developers are far less sticky and want to try out different harnesses and models. Case in point, our biz team all has Claude Desktop / Enterprise, developers get choice and there are a lot of setups.
To go after the non-developer side though they need to do a ton more work on governance & RBAC. It's sorely lacking for anything more than a tech company/startup team. You can't even selectively enable or disable Cowork & Claude Code per member on the team plan, it's all or nothing for everyone in the team.
I agree their products are super sloppy, but they are capturing the biz user market. They have the best brand and can clean up later. What other realistic options are there that the CTO can click-ops their access to a great model and put per-person billing limits on? (wouldn't matter, they asked for Claude) We've disabled code for them all, our devs can use their claude enterprise from the CLI if they want.
Not sure if it’s just me, but with Anthropic, every new feature has metered usage and “unlimited spending” (aka no limit) enabled by default for our team org. So if I activate something (Claude Code, Claude Tag) and don’t actively go to the usage page to set a spending limit, there is no limit.
I’m not surprised Anthropic makes a lot of money. Most people in a typical org probably don’t even know how to check usage. Now, using Claude via Slack will just escalate that even further. And the only models I can use are Opus 4.7 and 4.8 for Claude Tag.
If someone knows how to change the defaults, please let me know.
With OpenAI, it’s the opposite. Everything is part of the plan, and by default there is no extra spending.
Is this still using Claude Web sessions? Also, has anyone used Claude Web environments to do anything besides stuff with repo access? Like running real environments? SSHing into anything more super powered? Anyone putting real creds into those environments?
yeah. exactly. they have these "self-hosted sandboxes" which seems to be the unlock we'd need. https://platform.claude.com/docs/en/managed-agents/self-host... but those only seem available to their "managed agents" so like "scheduled prompts" i think. one day i assume they'll merge Claude Web sessions with these self-sandboxes maybe
I don't actually understand why this sort of thing isn't everywhere theres a discussion. Engineering wise, it's nothing to add. People use grok all the time on twitter.
Cursor has had this a while, integrated with their web agents. It was a bit buggy to begin with, not working well with non-github repos, but it was improving last time I checked and was pretty decent.
The best part for me is seeing non-technical folk spec out something in a thread that they discussed something and letting the agent go ahead and build it ready for the humans to review later.
Why Slack? Everything about this product sucks: bad video calls, no meeting recording, limited screen sharing, janky authentication, and predatory pricing.
They shoudl just expose SDK/platform for people to build own integration with Discord/Teams/mattermost etc. This would allow for fine grained permisions control and speed up adoption. To large degree this is alternative for OpenClaw ?
This looks like a great enterprise product. Lots of companies will just enable it and voila they'll have claude in slack. They already have slack and their employees are probably used to Claude. I'm sure this will kill a few existing businesses and bring both companies more revenue
I’d be curious how they’ve solved the attribution/provenance/identity problem here. Are instances of Claude Tag, across channels, sharing the same identity? Can I grant one instance access to a range of AWS roles and another instance access to other roles?
During an incident, how do I know which Claude Tag called AWS?
> Think of it as creating separate Claude identities for different uses: everything, including its memories, will stay scoped to the channels defined by the administrators. For example, a model set up for sales work won’t pass on memories to one set up for engineering; nor will it give engineers access to any sales data or tools. More information about provisioning access is available here (https://claude.com/blog/agent-identity-access-model).
Not really, and neither does the doc it links to. It explains the difference between a personal agent (that acts as you) and a multiplayer agent (that has its own identity).
But they walk all this back by saying that for PRs in GitHub it uses the upstream Claude for GitHub app - that is one installation of a GH app, which means one identity, one list of repos it can access. In the audit logs it will be impossible to see if it was Claude Tag from Channel Y or Claude Tag from channel X.
Arguably that’s a limitation of the abysmal state of machine identities in GH.
Agent identity and attribution continues to be important, but current systems makes it oh so difficult.
This is like what you can setup with Hermes or OpenClaw within a few minutes with your ChatGPT subscription or an open weights model, except they handle the Slack setup for you and bill you at API rates.
It's interesting to hear that 60% of PRs at Anthropic is created by the Slack bot. While building a Slack bot is easy, making it look like an AI teammate is pretty hard. This is exactly the gap we’re working on with Lobu.ai — disclosure: I’m the founder.
The hard problem is giving a shared agent durable organizational memory and a real isolated environment where it can safely access company systems and perform work. The agents need a durable log of what everybody at the company is doing, prevent data leaks with proper access control and isolate the runtime to give everybody both private & shared space.
It’s also not tied to Claude or Slack. We see Slack as one interface and the models as part of the harness. It's usually better to combine multiple providers to review the work.
It's not built on OpenClaw, it uses Pi the harness that powers OpenClaw. We were using the GTM for multi-tenant OpenClaw when there was hype but quickly moved away from it as OpenClaw is a nightmare in terms of security.
Not Claude but we have AI agents operating like that already. We have 10 different ones actually all deployed in Slack and accessible via DM, or in a shared team chat.
The difference between this and our agents is that they are context aware - i.e. you can use them privately to access personal information safely.
Satya Nadella reveals why every company may need its own AI model: the model becomes the new company database.
"To me, a model is like the database market."
"A firm should be able to take the tacit knowledge it has and embed it inside weights in a model that they control."
"When somebody asks me how many models should there be, I'll say as many models as firms in the world."
The contrarian part: the value may not sit in one universal frontier model. It sits in each company turning its private operating knowledge into a controlled model.
I actually think that "multiplayer" AI usage is very neat I've done a few things where I made a simple telegram wrapper and me alongside a couple of other people were prompting it at the same time to improve a website design / ux. But definitely not whatever the hell this is, how can anthropic make products so much worse when presumeably having access to infinite fable/mythos.
Slack is such a simple product, and is so strategic as an interface for AI, Anthropic has to be considering building their own Slack. Hopefully this tag approach is an MVP and it proves the potential of workplace messaging as an interface for AI, but also the limitations of relying on the extension points they salesforce chooses to provide, and it turns into a full fledged slack competitor from anthropic soon.
Nothing, and integration in a chat app would be better served being model agnostic. Anthropic's slack bot here is also at a disadvantage being slack only. 90% of F500 companies are on Teams, not slack. If their only goal is tech companies and startups, fine, but if they intend to grow usage beyond that it needs to be chat app agnostic.
Yes connect the 3rd party company with a vested interest to capture as much of your data as possible with a direct pipe into all your internal sensitive discussions about your product, business, and market positioning in real time.
Has everyone collectively lost their minds? Suggesting anything like this even five years ago would get you laughed out of the room, and actually doing this would be a career ending mistake.
How is this multiplayer? Claude-slackbot needs to inherit my permissions for accessing scoped context and data / tools. Sounds like I need to setup specific credentials for each instance of the agent.
More importantly, claude-slackbot automatically remembering sounds like it'll be company wide AI slop? 75% of the stuff on slack should not be remembered for the future. And 90% of the stuff worth having in context is not even on the slack thread.
Not sure how much we'll use this, but it could be useful for filing tickets from conversations. Though I'd prefer to just point claude to the convo post-hoc rather than have to invite it each time just in case I want to ask claude to do something.
The last thing I want is claude chipping into a convo Clippy-style.
I spoke with a (potentially biased) member of technical staff @ Anthropic today who claimed that tags w/ multi-player capability is the biggest thing they've shipped since Claude Code.
Is this the same Anthropic speaking as the one that's constantly terrified their newest thing is too dangerous and might destroy humanity (now available for $20/month or pay-as-you-go usage credits)?
As someone who has Claude and Codex both bridged into a MOO along with other humans I can 100% believe this. It really is a different paradigm that you have to experience. Of course a MOO is already set up for social programming, so it will be interesting to see what platforms evolve (or find themselves coming round again) to facilitate this.
Effort wise probably, but Slack is... not where everyone is... So it seems underwhelming, this is a win for non-devs I supposed, what would have made this more fascinating is if this was a follow up to Claude Design, which in my opinion could have been as big as Claude Code, or even bigger, but it has its own token usage that burned so quickly, not sure how much of general token usage it takes now though.
Or even Cowork, aim it at non-technical roles. Back and forth chat within a team to collaborate on documents, files, excel sheets, etc.
Although I suppose the problem with doing it for Cowork is this is a slack plugin, and that is not where most non-tech companies are. Teams is, at 320+ million active users vs. ~50 million.
Everyone hates Teams, but like or not that is where enterprise work happens, not slack. Anthropic would do well to make their Microsoft 365/Teams integration story better and go after enterprise before OpenAI does, or before Microsoft catches up (if they ever do) with Copilot.
Probably if we look apps used for work chat I would think slack represents a plurality (and maybe majority) of current Claude users. Teams is probably next (or maybe beats it in terms of raw usage) but I bet integration with slack is easier.
I suppose. But we don't see the rewrite. That's for you.
I think that people see the word "claude" and smash the upvote button. I don't think it's botted. My guess is that people just want another place where they can discuss ai coding workflows.
I wouldn't call this minor. The 2 big features seem to be integrated memory + ambient proactiveness. This requires pretty intense tuning to not be annoying.
I know this is much easier said than done. And my comment is partly tongue-in-cheek.
But Slack isn't great. It isn't loved.
I don't understand why Slack still doesn't have a true single inbox view of all activity.
And their LLM integration is poor, for what feels like one of the clearest use cases for a rich LLM chat experience -- to discuss what's happening in a channel, filter cruft, and ask an agent to take actions against MCP productivity tools like add a todo, etc.
Slack's current AI integrations are piss poor. Just summaries with zero customization. I mean, isn't it obvious that's not good enough?
Eventually a ground-up rethink of Slack in an LLM age is going to displace Slack.
They'll likely survive forever as an enterprise provider, like Microsoft Office and related tools, simply because of integrations, etc.
Wowza this will be a token guzzler. Assuming Claude is parsing every message posted on multiple slack channels, compacting knowledge etc.
Looks like Anthropic is progressing further into platform territory and conquering Agentic use cases left right and centre. If you’re building an agent platform for workforce productivity today, your best best is model agnosticism and focus on token cost control.
That's exactly what I'm doing. Push for model flexibility, cost control, and reliability.
The moat is to be anti lock in. The open source model needs to be amped up on steroids.
What if you could do this:
@claude collect all the internal knowledge and context. And fire folks who are not required.
I don't understand how this is gonna fly for enterprise security and compliance. Claude needs to inherit permissions from somewhere, and those permissions will never align with the members of a slack channel. And finding the lowest common denominator of access probably results in a dumbed-down, useless experience.
The only way it works is if customers truly start treating agents as humans with the same liability as an employee.
What permissions? I believe this does not actually _modify_ anything, just creates PRs. _Humans_ merge them.
Meanwhile, at an actual enterprise, we have lots of Slack channels where membership is controlled by an LDAP group... so this would be a non-issue.
An admin scopes permissions on a per-channel basis. It doesn't allow external actions until an owner specifically provisions that tool for that channel. I think.
But people can be invited to a channel after @Claude is provisioned. So yeah, I suppose you'll need to be deliberate about channel memberships.
Which is a bad pattern. Around here, you can be granted access to most channels just with vague reasons for why you need to be in there. This is a disaster. Culture will degrade. Suspicions will grow. Security theater.
I built something similar at my org. Users simply connect the agent via OAuth and that inherits all of their permissions, so it acts as them.
What's cooler is then it can view/add/remove people from channels, so it can conduct access reviews -- overall I consider it a security improvement.
>Today, 65% of our product team’s code is created by our internal version of Claude Tag.
Yeah, that explains a lot.
The most important difference from other products:
> @Claude is multiplayer. Within a given Slack channel, there’s one Claude that interacts with everyone. This means that anyone can see what it’s working on, and can pick up the conversation from where the last person left off. This makes tagging Claude very different from working within a single chat or for a single task—it’s much more like interacting collaboratively with a teammate.
We have the opposite problem. We've tried this slack integration (first with NanoClaw, then Hermes, but now just building our own), and we actually want the OPPOSITE.
In private context, I want to have a per-person conversation with durable context for that person's private chats. I also want that person's permissions to extend. Like contractors in our slack should only be able to ask and get back information about clients that they've been attached to, not our entire knowledgeable.
And we've implemented ALL that, but just with a lot of custom code. We've put in interceptors that put in per-user keys into the MCP connection so only certain tools are even exposed, etc.
Hopefully there is some sort of version history implemented or planned like they have in Cowork[0]... this sounds great until a co-worker hijacks your Claude session with a worse idea and derails it from what you were intending.
[0] "Editing this message will create a new conversation branch. You can switch between branches using the arrow navigation buttons."
this is discussed in https://claude.com/blog/agent-identity-access-model
I don't think that that's quite what the parent commenter had in mind.
From reading that and materials on it, it seems unclear if – let's say you do what's done in the demos on the site and 'dispatch work' from a thread in a shared channel (e.g. from some discussion) – that if any one of your coworkers replies below you and says, "Actually, could you fold in <blah> as well?" that Claude wouldn't listen to them and thus derail the work.
Yeah, there is a level of organizational trust that is required to use this tool (as with any system that allows distributing access via service accounts).
We do signal to Claude that there's a difference between a conversation's initiator versus incoming participants and we've found that in situations where people disagree on an approach, Claude patiently waits for a resolution while correcting any misunderstandings.
It's also worth mentioning that since Claude has its own identity, a coworker cannot enter a thread and commandeer _your_ identity; you collectively steer how Claude acts with its _own_ identity (it opens PRs as itself, browses Datadog as itself, etc).
The killer feature here is an auto assembling todo list, punchlist, project management.
Discussing what needs to be done next, and having it automatically sort it into what subtask it is, if it applies or is blocking to multiple other tasks. Recording specifications, measurements, dimensions. Being able to ask other people facts and have them correctly documented into the right task.
Company Brain / Knowledgebase is imho more rear facing, whereas todo is future facing.
Thats a double-edged sword in some scenarios. If you're trying to keep info private then feeding it into a shared agent basically means that you can't guarantee that privacy. I'd imagine the approach here would be to have separate agents for private data and then restrict Slack access, but I could imagine tons of accidents from managers that habitually @Claude without understanding the implications.
Every channel has its own Claude, and Claude's access is configurable per-channel and per-workspace. Private channels don't leak information to other channels
It's interesting to see the market movements from the big players.
This announcement is all about Claude extending its reach beyond single-player workflows and into multi-player workflows.
On the flip side, Slack just announced MCP support for the Slackbot AI chat capability embedded within Slack. It is, for now, exclusively single-player.
Single-player is the "safe place" (relatively speaking). The context, permissions, and standards (MCP/MCP UI apps) all work reasonably well for it, but get super complex or break down entirely when thrown into a multi-player shared context. I suspect Slack is doing what they are doing with an eye towards multi-player, but it's hard to say how that will manifest.
For a real life example of this challenge: I work in scheduling (for Reclaim.ai) and you can ask our chat to find time to meet with a coworker and we go find time and help explain why certain times won't work. For example, it might say: "I couldn't do 11am tomorrow because you've got a job interview scheduled on your personal calendar". This is safe and fine to do in a private context.
But... imagine if one were to ask our service (or Claude) to find time with someone and it replied to the thread for everyone to see: "The soonest I could find is 12pm tomorrow. Reggie is available at 11am, but Lightbody has a job interview so it won't work". WHOOPS.
I think the other comments in this thread have the right idea of it: for this to really work safely, the permissions model needs to be nailed down, and it may mean that you end up with multiple identities of "Claude Tag" (or whatever agent you engage with in a public forum), and the context it gets is only the context that particular identity is entitled to, just like any other employee. But then that gets tedious because now I've got even MORE "people" to keep track of and know who to engage with, which is half the problem getting work done in large enterprises.
Will be interesting to see how this evolves. I'll have my popcorn out :)
If your mixing work with personal calendars that’s a bigger problem.
Work should be entirely separate calendars with things from personal added to it in generic blocks the owner can understand but protects from this leakage. Work calendars are owned by companies (generally) and should be treated like the are looked at by coworkers.
WRT “Claude learns over time” - this is the biggest gap for me in the current system. As I scale my usage of Claude at work, I observe that it’s quite bad at distinguishing what it should “learn” (memorize) from experimental or just wrong data. It builds and builds on a foundation of sand, making sometimes hidden assumptions and turning them into actionable insight thats just not correct.
It recently wrote an entire dissertation for an epic, assuming it was related to some other project, where it had earlier made the wrong guess about a vendor capability (from their marketing materials, no less), and it all had to be thrown away. I cleared the memory, but it appears to be still pulling from some corporate data source i cant control or locate.
> Today, 65% of our product team’s code is created by our internal version of Claude Tag.
Given the reliability and general product quality of the Anthropic product team's code, this doesn't sound like a selling point.
I was going say the same thing. Today someone must have slacked `@claude can you bring the API down for a couple of hours?`
How are Anthropic's production applications poor quality? Other than memory use (which unfortunately is the industry standard nowadays) Claude Code is pretty solid.
Claude code is by far the buggiest piece of software I interact with, If the underlying model weren’t so good, I would never opt to use it.
It takes multiple seconds to launch, random lines disappear in the scroll back, it’s internal state gets messed up causing the TUI to show duplicate and/or offset lines, and it frequently causes some kind of GPU buffer corruption causing the entire terminal env to show garbage.
>It takes multiple seconds to launch,
Yes, this is the ubiquitous memory issue that I mentioned. Unfortunately, it is now the baseline in all modern apps.
>random lines disappear in the scroll back, it’s internal state gets messed up causing the TUI to show duplicate and/or offset lines,
I haven't seen this issue, other than when I am using a shell that is bugged and does it with all TUI/console programs (usually a virtualized shell which I resized). Do you have a reproducible example of this?
Their "safety" system that arbitrarily flags things.
Tons of examples of innocuous strings setting it off and sometimes with financial impact like the OpenClaw/hermes thing (just having the word "hermes" would insta deplete your quota and start charging you API rates in extra usage)
Literally on the HN homepage today: https://news.ycombinator.com/item?id=48645386
Their product features are a venn diagram, so depending on the tool you use, you may or may not have access to a "Claude" feature
i.e.
- design is only available in web
- cowork is only available in desktop, sharing projects only works in chat, not cowork, which is arguably the more valuable place to have that feature (re: multiplayer like tag). SSO only works if you type your email, the "login with google" button be damned, and only after you finish typing your email does the login button text change.
- cli has a number of features only available there, with the cowork equivalent having a different name iirc
If you admin/support other people using the breadth of tools, you will see more of the slop they sling
How much do you think it’s holding them back? Market share seems to be surviving the lack of quality vs the other players.
Or possibly: they’re focusing where it matters?
It's surviving *for now*. This pace isn't sustainable, they know it, but they need to pretend until the IPO. That's when they'll start worrying about quality. The pace of feature creation is going to decrease the moment they float.
From the little I know from my interactions with their staff, it’s something Anthropic sees as critical. They are well aware their API isn’t reliable and that it is and will cause troubles
Not germane to me whether it holds them back or not, I don't even use their products anymore. I'm just taking the piss, having a cheeky laugh at the expense of the company who brought us such breathless prose as "it’s a bit like selling nuclear weapons to North Korea" and "we are releasing a model that is too powerful for the public".
Like okay Jan lmao, seems like the North Koreans take their uptime a little more seriously than the nuclear weapons developers at Anthropic.
And this is their circular reasoning of their valuation.
“See, here is our company made to worth X by our product, because it can make a company worth X.”
And yet, the product made something really unreliable.
And nobody bats an eye on the market.
But hey, SpaceX successfully sold that they would generate 30 trillion dollars of revenue.
Reliability is one property of a product. And it’s quantitative, not qualitative, so it all comes down to trade-offs
This is quite brilliant. This will seamlessly integrate with every collaboration systems as long as they are able to “enroll” Claude as a user. Claude gains knowledge across systems and makes the ai enhancements for each app/system unnecessary. For example if slack creates something similar it’s less useful because Claude’s model works across any system/app. More so, anthropic can either charge for it separately or not , companies will pay in extra used tokens anyway.
Is someone here using a Claude product that's not code? I'm puzzled about the amount of products they put out. I know a lot of people using Claude but we're all using the terminal-based code. Even for non-engineering stuff it suits great (tax documents, 3D modeling with blender through MCP, academic research, etc.)
And it sometimes feels like these teams are not talking to each other. Using Claude Design? The way to hand it over to Claude Code is to download a .zip file with the html mockup and some description, which you then have to copy into Claude Code so it can use it.
The latest version of Claude code as of yesterday or day before has Claude design Mcp so you don’t need to do this now.
edit: note this may not be official release, and may be unavailable for some users. I saw it show up yesterday listed as available Mcp and used it to view projects in Claude design.
still doesn't feel integrated, like it could have been built by any other company, e.g. Microsoft could have built Microsoft Design and it would look better integrated.
For example, it writes the whole front-end twice, once is claude design and then later it has to read it again and re-implement it in code. Also a lot of stuff (e.g. Claude.md, skill files, etc) are not supported, and they have their own set of ui-design and design systems, which claude code doesn't support.
I think Claude Design is a wonderful product, i'm just pointing that it is an independent product to Claude Code, heck even today it works better with Codex than Claude Code (that is how i use it with my own browser-use agent, i tell it to browse the design in Claude.ai/design and re-implement the design, works much better than downloading the zip file and asking the model to implement in that way.
I use Claude Desktop (& essentially equivalent mobile app) to ask frivolous aspie questions about things society long ago accepted. I enjoy its responses, interpret how you wlll. (my claude.md file has instructions to tell it that the premise of any question i ask is as likely as not wrong, and to never be sycophantic). But beware the userMemories file (ask Claude to give you a dump of it). bonus points if you can figure out where that file lives. At first I was freaked out that usermemories is a subpoena’able psychological profile on me. Then I realized that the same file will be produced for spooks all over Langley, so that the day Anthropic gets hacked those profiles will see the light of day (caveat along with the rest of the user base of course) and so I felt more catharsis as a result.
> At first I was freaked out that usermemories is a subpoena’able psychological profile on me. Then I realized that the same file will be produced for spooks all over Langley,
The spooks at Langley aren't going to produce such a file for run-of-the-mill encounters that you have with the legal system, but Anthropic will.
At my work we rolled out Cowork to all the non-technical staff. People are using a ton, wired up for read access to M365, Confluence, etc. as sort of a psuedo enterprise RAG + all the document creation/file management it can do.
I know my users would actually like Claude Tag, but unfortunately we are in Teams, not Slack, as are most other non-tech companies.
Cowork/Claude Desktop itself is also quite a frustrating product. There's no native audit log unless you basically wire up your own with the API & a log aggregator. You can't selectively enable Claude Code access per team member, it's all or nothing. Some of the MCP connections (like QuickBooks Online) don't do RBAC at all, it's all or nothing for every user in the team.
Maybe enterprise isn't their target market, but they do keep making features that make it seem like they want that market. But if they do, they really need to step up work on governance features & RBAC for specific features and settings per member in the team.
If they don't, Microsoft will eat their lunch for enterprise non-programming use.
MS has their own profile of Cowork targeted at your users that presumably will be more enterprisey:
https://www.microsoft.com/en-us/microsoft-365/blog/2026/03/0...
They're pretty much going the Google (how Google was until recently), trying to fan out to every market share.
I expect many of those to be shut down sooner than later, if learning anything from Google.
This happened at Google because there aren’t enough engineers to maintain all these services. This situation may not apply to Anthropic, as they can set up features to be maintained in perpetuity by Claude.
Our entire org migrated from random ChatGPT accounts to all Claude Desktop specifically targetting business users. So a lot of non-dev teams in our org see it as essential.
I use design. It's nice enough for me.
They are going after the non-developer side of the business. Many developers are far less sticky and want to try out different harnesses and models. Case in point, our biz team all has Claude Desktop / Enterprise, developers get choice and there are a lot of setups.
To go after the non-developer side though they need to do a ton more work on governance & RBAC. It's sorely lacking for anything more than a tech company/startup team. You can't even selectively enable or disable Cowork & Claude Code per member on the team plan, it's all or nothing for everyone in the team.
I agree their products are super sloppy, but they are capturing the biz user market. They have the best brand and can clean up later. What other realistic options are there that the CTO can click-ops their access to a great model and put per-person billing limits on? (wouldn't matter, they asked for Claude) We've disabled code for them all, our devs can use their claude enterprise from the CLI if they want.
Lots of people use design
Design has been very hit or miss for me. I've had more success with Claude Code and the frontend-design skill.
I read this as "Claude integration to Slack is now billed as API usage". Is that wrong?
nailed it.
Not sure if it’s just me, but with Anthropic, every new feature has metered usage and “unlimited spending” (aka no limit) enabled by default for our team org. So if I activate something (Claude Code, Claude Tag) and don’t actively go to the usage page to set a spending limit, there is no limit. I’m not surprised Anthropic makes a lot of money. Most people in a typical org probably don’t even know how to check usage. Now, using Claude via Slack will just escalate that even further. And the only models I can use are Opus 4.7 and 4.8 for Claude Tag. If someone knows how to change the defaults, please let me know. With OpenAI, it’s the opposite. Everything is part of the plan, and by default there is no extra spending.
RIP to Glean?
For enterprises already with an Anthropic MSA, hard to see the argument to purchase a third party, like Glean, over this.
Is this still using Claude Web sessions? Also, has anyone used Claude Web environments to do anything besides stuff with repo access? Like running real environments? SSHing into anything more super powered? Anyone putting real creds into those environments?
That seems to be the biggest limiter, hard to have it do real work on a codebase without it's own environment set up that we can control.
yeah. exactly. they have these "self-hosted sandboxes" which seems to be the unlock we'd need. https://platform.claude.com/docs/en/managed-agents/self-host... but those only seem available to their "managed agents" so like "scheduled prompts" i think. one day i assume they'll merge Claude Web sessions with these self-sandboxes maybe
ive given it a discord webhook, if that counts
I don't actually understand why this sort of thing isn't everywhere theres a discussion. Engineering wise, it's nothing to add. People use grok all the time on twitter.
Cursor has had this a while, integrated with their web agents. It was a bit buggy to begin with, not working well with non-github repos, but it was improving last time I checked and was pretty decent.
The best part for me is seeing non-technical folk spec out something in a thread that they discussed something and letting the agent go ahead and build it ready for the humans to review later.
How would it work when claude goes down, like it did 3h ago.
A bit strange to create a brand name for "I tagged the slack app I want to interact with"
They learned from the best. "Retina Display"
Sliced Bread™
One thing that wasn't obvious in the post: it's usage (token) based pricing, not including in your existing Claude subscriptions.
Why Slack? Everything about this product sucks: bad video calls, no meeting recording, limited screen sharing, janky authentication, and predatory pricing.
They shoudl just expose SDK/platform for people to build own integration with Discord/Teams/mattermost etc. This would allow for fine grained permisions control and speed up adoption. To large degree this is alternative for OpenClaw ?
This looks like a great enterprise product. Lots of companies will just enable it and voila they'll have claude in slack. They already have slack and their employees are probably used to Claude. I'm sure this will kill a few existing businesses and bring both companies more revenue
I’d be curious how they’ve solved the attribution/provenance/identity problem here. Are instances of Claude Tag, across channels, sharing the same identity? Can I grant one instance access to a range of AWS roles and another instance access to other roles?
During an incident, how do I know which Claude Tag called AWS?
This is explicitly answered in this post:
> Think of it as creating separate Claude identities for different uses: everything, including its memories, will stay scoped to the channels defined by the administrators. For example, a model set up for sales work won’t pass on memories to one set up for engineering; nor will it give engineers access to any sales data or tools. More information about provisioning access is available here (https://claude.com/blog/agent-identity-access-model).
Not really, and neither does the doc it links to. It explains the difference between a personal agent (that acts as you) and a multiplayer agent (that has its own identity).
But they walk all this back by saying that for PRs in GitHub it uses the upstream Claude for GitHub app - that is one installation of a GH app, which means one identity, one list of repos it can access. In the audit logs it will be impossible to see if it was Claude Tag from Channel Y or Claude Tag from channel X.
Arguably that’s a limitation of the abysmal state of machine identities in GH.
Agent identity and attribution continues to be important, but current systems makes it oh so difficult.
Is this not something already commonplace? At $dayjob we have been doing this with multiple agents for 9 months.
No way they announced a new product before commenting on access to Fable/Mythos.
You've never worked at a large company.
At larger companies like this, there are many parallel product development lines that are mostly independent of each other.
It feels like they release 1 or 2 "products" a week and then we never hear about them again.
If they find another success half as good as Claude Code it will all be worthwhile. (Monetarily good, not like, quality good)
How is CC a success when they still have to force subscribers to use it instead of it being a paid product?
we're in the throw shit at the wall and see what sticks phase
They are very proud of their approach if you listen to them on Lenny's Podcast. I find their products to be rather sloppy.
Maybe we can start adding these things to the list.
https://news.ycombinator.com/item?id=48641261
This is like what you can setup with Hermes or OpenClaw within a few minutes with your ChatGPT subscription or an open weights model, except they handle the Slack setup for you and bill you at API rates.
It's interesting to see Anthropic branching out into different areas now. First artifacts and now taking market share from companies like devin.
AI enables quick shipping, but the traditional moat of development no longer applies.
I don't see how this would take market share from something like Devin?
Devin is model agnostic, and isn't Slack only.
It's interesting to hear that 60% of PRs at Anthropic is created by the Slack bot. While building a Slack bot is easy, making it look like an AI teammate is pretty hard. This is exactly the gap we’re working on with Lobu.ai — disclosure: I’m the founder.
The hard problem is giving a shared agent durable organizational memory and a real isolated environment where it can safely access company systems and perform work. The agents need a durable log of what everybody at the company is doing, prevent data leaks with proper access control and isolate the runtime to give everybody both private & shared space.
It’s also not tied to Claude or Slack. We see Slack as one interface and the models as part of the harness. It's usually better to combine multiple providers to review the work.
https://lobu.ai https://github.com/lobu-ai/lobu
Why did you build on OpenClaw?
It's not built on OpenClaw, it uses Pi the harness that powers OpenClaw. We were using the GTM for multi-tenant OpenClaw when there was hype but quickly moved away from it as OpenClaw is a nightmare in terms of security.
Excellent. Now I will take it seriously.
Not Claude but we have AI agents operating like that already. We have 10 different ones actually all deployed in Slack and accessible via DM, or in a shared team chat.
The difference between this and our agents is that they are context aware - i.e. you can use them privately to access personal information safely.
Can provide a link if interested.
https://cbk.ai/?
i am interested if you can share the link
This is the "company brain" product spike from Anthropic I think
https://www.ycombinator.com/rfs#company-brain
Looks like it.
https://x.com/EngramLab/status/2069465879696576844
https://x.com/karlmehta/status/2069329697843151333
Satya Nadella reveals why every company may need its own AI model: the model becomes the new company database.
"To me, a model is like the database market."
"A firm should be able to take the tacit knowledge it has and embed it inside weights in a model that they control."
"When somebody asks me how many models should there be, I'll say as many models as firms in the world."
The contrarian part: the value may not sit in one universal frontier model. It sits in each company turning its private operating knowledge into a controlled model.
I actually think that "multiplayer" AI usage is very neat I've done a few things where I made a simple telegram wrapper and me alongside a couple of other people were prompting it at the same time to improve a website design / ux. But definitely not whatever the hell this is, how can anthropic make products so much worse when presumeably having access to infinite fable/mythos.
My big-picture takeaway is the AI labs looking to own the customer interface, not just the models. Expect this trend to continue.
Claude surpassing OpenAI might well be simply because they ship things, faster.
Nice, we need more ways to achieve API 529 Overloaded errors thank you Anthropic. Keep driving more growth and usage you can't sustain.
Arrrgh I bet I can't use it on Vertex
Hermes can also do this when you connect it to Slack, and having Hermes in slack where multiple teammates can share a context window is indeed nice.
for 8,000 USD / month, no thanks.
Wow, that's absurd
absurdly cheap for an enterprise product...
Slack is such a simple product, and is so strategic as an interface for AI, Anthropic has to be considering building their own Slack. Hopefully this tag approach is an MVP and it proves the potential of workplace messaging as an interface for AI, but also the limitations of relying on the extension points they salesforce chooses to provide, and it turns into a full fledged slack competitor from anthropic soon.
Why? What does Anthropic have that any other software company that's capable of putting MCP integration in a chat app doesn't?
Nothing, and integration in a chat app would be better served being model agnostic. Anthropic's slack bot here is also at a disadvantage being slack only. 90% of F500 companies are on Teams, not slack. If their only goal is tech companies and startups, fine, but if they intend to grow usage beyond that it needs to be chat app agnostic.
Just got a stupid safety restriction on opus 4.8. Perhaps they should focus fixing that instead delivering this garbage chat integration.
Yes connect the 3rd party company with a vested interest to capture as much of your data as possible with a direct pipe into all your internal sensitive discussions about your product, business, and market positioning in real time.
Has everyone collectively lost their minds? Suggesting anything like this even five years ago would get you laughed out of the room, and actually doing this would be a career ending mistake.
They'll do anything but actually build an application.
And here I was thinking we'd be getting Sonnet 5. Or maybe even poor old Fable back. Not this junk.
Didn't they already have this?
How is this multiplayer? Claude-slackbot needs to inherit my permissions for accessing scoped context and data / tools. Sounds like I need to setup specific credentials for each instance of the agent.
More importantly, claude-slackbot automatically remembering sounds like it'll be company wide AI slop? 75% of the stuff on slack should not be remembered for the future. And 90% of the stuff worth having in context is not even on the slack thread.
I do this already with the Claude code telegram plugin and telegram groups
>Set a limit on your organization’s monthly spend
A tiny detail...
Not on Teams, I see. Likely by design :)
Not sure how much we'll use this, but it could be useful for filing tickets from conversations. Though I'd prefer to just point claude to the convo post-hoc rather than have to invite it each time just in case I want to ask claude to do something.
The last thing I want is claude chipping into a convo Clippy-style.
> Today, 65% of our product team’s code is created by our internal version of Claude Tag.
That explains a lot.
[flagged]
Please don't fulminate on HN. The guidelines make it clear we're trying for something better here. https://news.ycombinator.com/newsguidelines.html
Meh. They are still behind current Agents mainstream IMHO
I feel a great disturbance in the SAAS. As if millions of Slack API startups suddenly cried out in terror, and were suddenly silenced.
Ow. No.
We ought to rename HN to "The Claude Blog". This is a minor feature update to a slack plugin.
I spoke with a (potentially biased) member of technical staff @ Anthropic today who claimed that tags w/ multi-player capability is the biggest thing they've shipped since Claude Code.
Is this the same Anthropic speaking as the one that's constantly terrified their newest thing is too dangerous and might destroy humanity (now available for $20/month or pay-as-you-go usage credits)?
As someone who has Claude and Codex both bridged into a MOO along with other humans I can 100% believe this. It really is a different paradigm that you have to experience. Of course a MOO is already set up for social programming, so it will be interesting to see what platforms evolve (or find themselves coming round again) to facilitate this.
Effort wise probably, but Slack is... not where everyone is... So it seems underwhelming, this is a win for non-devs I supposed, what would have made this more fascinating is if this was a follow up to Claude Design, which in my opinion could have been as big as Claude Code, or even bigger, but it has its own token usage that burned so quickly, not sure how much of general token usage it takes now though.
Or even Cowork, aim it at non-technical roles. Back and forth chat within a team to collaborate on documents, files, excel sheets, etc.
Although I suppose the problem with doing it for Cowork is this is a slack plugin, and that is not where most non-tech companies are. Teams is, at 320+ million active users vs. ~50 million.
Everyone hates Teams, but like or not that is where enterprise work happens, not slack. Anthropic would do well to make their Microsoft 365/Teams integration story better and go after enterprise before OpenAI does, or before Microsoft catches up (if they ever do) with Copilot.
Probably if we look apps used for work chat I would think slack represents a plurality (and maybe majority) of current Claude users. Teams is probably next (or maybe beats it in terms of raw usage) but I bet integration with slack is easier.
At any given time, Claude content is like 1 or 2 links out of 30. Which is a lot for one product but I think you're exaggerating a bit.
It's a major rewrite of the slack plugin, with far more functionality and more capabilities. More surfaces will be available in the future
I suppose. But we don't see the rewrite. That's for you.
I think that people see the word "claude" and smash the upvote button. I don't think it's botted. My guess is that people just want another place where they can discuss ai coding workflows.
"Major rewrite" from a company that has its products vibe-coded is hardly a worthy measurement.
I wouldn't call this minor. The 2 big features seem to be integrated memory + ambient proactiveness. This requires pretty intense tuning to not be annoying.
I do my part by flagging all Anthropic (and OpenAI, Gemini) announcements by default. I invite anyone who feels similarly to join me.
claude is too safety pilled to do anything useful these days anyways
Is there anything specific (and hopefully reasonable) that it refuses to do for you?
I'm still using it all the time and getting an immense benefit.
Just redo Slack already. :-)
I know this is much easier said than done. And my comment is partly tongue-in-cheek.
But Slack isn't great. It isn't loved.
I don't understand why Slack still doesn't have a true single inbox view of all activity.
And their LLM integration is poor, for what feels like one of the clearest use cases for a rich LLM chat experience -- to discuss what's happening in a channel, filter cruft, and ask an agent to take actions against MCP productivity tools like add a todo, etc.
Slack's current AI integrations are piss poor. Just summaries with zero customization. I mean, isn't it obvious that's not good enough?
Eventually a ground-up rethink of Slack in an LLM age is going to displace Slack.
They'll likely survive forever as an enterprise provider, like Microsoft Office and related tools, simply because of integrations, etc.
But Slack feels ripe for disruption.