# Day 46 — YouTube 自动生成字幕（AUTO_GENERATED）

覆盖训练段 000–10 分钟。已去滚动字幕重叠。ASR 生成可能含识别错误。

==================================================

And so we think of Slack as the best home for basically employee agents, the ones that you make available employees. And we think of Slackbot as the employee super agent that can help make [music] handoffs to all these other agents. It's a personal agent. It's for you. It's not for teams. It's not for entire it's not for for the company. It's for you. We went from zero to like 40,000 weekly active users in a matter of a few weeks. >> I do think in the next

couple years you'll see as many agents in Slack as you do humans. I think our old our old ways of thinking about capital expenditures, expenditures, of building business cases, like that stuff is completely invalid right now and [music] and [music] >> more important that you try things as fast as you possibly can. >> Hello and welcome to another episode of our own devices. Uh we are going to talk about a product um which a lot of us use almost every day. We can't we can't live

without it. Uh and it's a product that's evolving in the interesting times that we are in. Uh and we're going to talk about uh Slack and Slackbot which is their new product. It's just been announced. Uh and we have with us Rob Seaman who is the chief product officer and CEO of Slack. So Rob, thanks for being on the show. >> My pleasure. Thank you so much for having me and thank you to all the listeners that are customers and users of Slack. of Slack. >>

Yeah. So Rob, you have just announced Slackbot. Could you could you sort of you know tell our users in this AI world where we are all going agentic and we think agents are going to take over our jobs and also make life much easier for us. So so how do you think the Slackbot sort of you know fits into this new world we are in? So >> if you go back to um Slack for a long time has had Slackbot. it's actually been there for for

10 years or so and it's a it's a very rudimentary like notification system and it tells you're added to channels or you're removed from channels or channels been archived. >> Um so about a year ago we said what if we took this concept it's already there in your Slack and and people actually like have some endearment uh for for Slackbot already and turned it into a personal AI agent that could help you. And so we started prototyping and it became immediately clear that there was something

very powerful and lovable there frankly. And uh so I I think that the key thing with Slackbot is it's a it's a personal agent. It's for you. It's not for teams. It's not for entire it's not for for the company. It's for you, you know, and every single individual because it has access to your Slack. Um meaning it was what's publicly available in your Slack. It can access the private channels and direct messages that you're part of for you. It can also access the systems that

you have connected to your Slack. Um, so it's become, you know, in our in our early dog fooding of our own use, the roll out to Salesforce and with 50 plus pilot customers, it's really become an indispensable personal agent and people love it. It's it's really cool to see. We've seen over 80% week-over-week retention of users using it. So um you know in my last season of the podcast I had a chat with Mohawk Shro who's the chief product officer at LinkedIn and and he gave

me this catchphrase that the agentic AI is your chief of staff right you know so is the Slackbot going to be the chief of staff for you working inside your Slack ecosystem ecosystem >> I I think it is um and I think the key thing is it is meant to be a chief of staff for you um but it's also meant to be complimentary to your other AI tools. I want to be very clear that like we remain an open open ecosystem and we we want

our customers and users to bring the the products into Slack that that are the best fit for them. But we absolutely think of Slackbot as your chief of staff. Um it's like I use it I I use a great example. I used it last week to go look for the upcoming all hands presentation, find all of our new hires in the presentation and tell me how to pronounce their names. you know that's absolutely something a chief of staff would have done before you know and I

and it was able to do it in 10 seconds you know and uh it's it's pretty amazing amazing >> so you know I would say over the past few months the the adoption of agentic tools is also going up know it's not maybe scaled up the way people thought it would happen like you know but but everybody's now figuring out a lot of new things especially this week we're seeing very interesting products come out um so so Are enterprises getting the hang of it? Because you

know initially they suddenly had the AI boom then they had gen AI use cases they had new use cases coming up now agentic use cases are coming up. So so are enterprises getting the hang of it um you know are they having to pivot their understanding of what to do with this uh you know frequently. So so how are you looking at this entire scenario? what is your advice maybe to enterprises on how to adopt it? to adopt it? >> Yeah. Um so I think they

I think they are figuring it out but I do think it's a process of trial and error. There's there's no methodology and I I think that's okay. Um at this point I think I my my advice we we do this internally at Slack. My advice to customers would be to try as many things as you like fiscally responsibly can try. Um, and then see what works for you, your company, and your individual users. Um, we have tried to make Slackbot something that will work for all

of your users. And we think it will because it's close to where they're talking to their colleagues. It's connected to your Slack and their system. So, it's going to be very relatable. It's it's easy to find, very relatable, easy to use. Um, but we experiment with all kinds of other AI products. And I think what Slackbot can ultimately help do is when you find the things that um are meaningful for your company and you deploy them into Slack so that people can use them, Slackbot will

make referrals over to that and actually help people discover the AI that matters to you as a company and get them to use it, you know, and so I think that's the that's the role that we can play. But I would tell everybody just go try as many things as you can, you know, rapidly. I I I think our old our old ways of thinking about capital expenditures, expenditures, of building business cases, like that stuff is completely invalid right now and and >> more important that

you try things as fast as you possibly can because the the rate of change is incredibly fast right now. And if you're betting on business cases and long evaluations, like you're just going to take too long and you'll have missed the cycle, you know? So, >> yeah. So um also I know you spoke about agentic enterprise like you know so so how so you mentioned slackbot as being you know sort of like your personal assistant but but how is the agentic AI piece helping enterprises you

know take care of a lot of things right like you know you can you can build in a lot of you know um as in regular functions being sort of hived out to the agentic enterprise to take care of is that how you would look at it and um and then your employees themselves working on really deep stuff. >> Um I I'd say it's a it's a combination. So I'll talk a little bit about like what we've seen amongst our customer base and what we're doing

ourselves within Slack. So I think Slackbot gives you basically a personal assistant that you could chief of staff as you said to make available to every employee and that's going to be a productivity improvement for every employee. you mean it's going to save them time and give them time to work on higher order tasks, high higher value tasks and and be more creative. Now, the agents that we're we're seeing thousands of agents that are being deployed into Slack and that's everything from uh coding agents uh

to agents that are taking care of things that used to be handled through workflows um to new agentic workflow systems. Um but I think the probably the best example use case is actually a coding agent. And so we we've got a number of those that have been deployed to Slack from Verscell to Cloud Code to OpenAI's codeex. You can use all of those in in Slack now. But if you think about um where the conversation is happening about what you want to build or about the

changes you want to make to something you've already built, like it's typically happening in Slack channels, right? um it's happening in team channels or an individual feature channel that where you're saying hey you know this part of our our own devices website we should actually update the copy the logo over here do that and people are talking about that in channels well these agents can just be another member of the channel now they can listen can listen >> and you can say hey cloud code just

go ahead and look at look above here and see what we talked about and go make that happen or codeex go do that you know or burcell go do that and I think that's um probably one of the first places many companies companies can start and achieve a ton of value is is honestly from the coding agents but we're seeing it you know we're happen you see it happening in design and other areas as well areas as well >> so the way um you know the

way a lot of companies have have integrated Slack into their workflow it's an integral part it's like your spine uh so so is that a great place uh in a way for the agent to sort of rest because it gets the entire context of the organization because exactly what you said the conversations are all there right so it has the context it knows what is maybe taking the mind space of the company at that point like you know so so so how are you looking at

that so you know I hear a lot about uh orchestrating um uh these different agents but um

# Day 46 — 火山引擎 录音文件识别（volc.seedasr.auc）

豆包大模型 ASR，异步接口。可能含识别错误，仅供听力对照。

==================================================

So we think of Slack as the best home for basically employee agents, the ones that you make available for employees.

And we think of Slackbot as the employee super agent that can help make hand UPS to all these other agents.

It's a personal agent.

It's for you.

It's not for teams.

It's not for it's not for it for the company.

It's for you.

We went from zero to like 40,000 weekly active users in a matter of a few weeks.

I do think in the next couple of years, you'll see as many agents in Slack as you do humans.

I think our old our old ways of thinking about capital expenditures of building business cases like that stuff is completely invalid right now.

And more important that you try things as fast as you possibly can.

Hello and welcome to another episode of our own devices.

We are going to talk about a product which a lot of us use almost every day.

We can't, we can't live without it.

And it's a product that's evolving in the interesting times that we are in.

We're going to talk about Slack and Slackbot, which is their new product has just been announced.

Uh, and we have with us Rob Seaman, who is the Chief Product Officer and CEO of Slack.

So Rob, thanks for being on the show.

My pleasure.

Thank you so much for having me.

And thank you to all the listeners that are customers and users of Slack.

Yeah.

So Rob, you have just announced Slackbot.

Could you could you sort of, you know, tell our users in this AI world where we are all going agentic and we think agents are going to take over our jobs and also make life much easier for us.

So.

So how do you think the Slackbot sort of fits into this new world we are in?

If you go back to Slack for a long time has had Slackbot.

It's actually been there for for 10 years or so.

And it's a it's a very rudimentary like notification system and it tells you're added the channels or you're removed from channels or channels been archived um, so about a year ago we said what if we took this concept it's already there in your Slack and and people actually like have some endearment.

Uh for for Slack bot already and turned it into a personal AI agent that could help you.

And so we started prototyping and it became immediately clear that there was something very powerful and lovable there, frankly.

And so I think that the key thing with Slackbot is it's a it's a personal agent.

It's for you.

It's not for teams.

It's not for it's not for it for the company.

It's for you, you know, and every single individual because it has access to your Slack.

Um, meaning it?

Yeah.

What's publicly available in your Slack, it can access the private channels and direct messages that you're part of.

For you, it can also access the systems that you have connected to your Slack.

Um, so it's become, you know, in our in our early dogfooding of our own use, the rollout to Salesforce and with 50+ pilot customers, it's really become an indispensable personal agent and people love it.

It's it's really cool to see.

We've seen over 80% week over week retention of users using it.

So, um, you know, in my last season of the podcast, I had a chat with Mohak Shroff, who is the chief product officer at LinkedIn and.

And he gave me this catchphrase that the agentic AI is your chief of staff, right.

You know, so is the Slackbot going to be the chief of staff for you working inside your Slack ecosystem?

I I think it is.

Um and I think the key thing is it is meant to be a chief of staff for you.

Um, but it's also meant to be complementary to your other AI tools.

I want to be very clear that like we remain an open e open ecosystem and we, we want our customers and users to bring the, the products into Slack that that are the best fit for them.

But we absolutely think of Slackbot as your chief of staff.

Um, it's like, I use it, I, I use a great example.

I used it last week to go look for the upcoming all hands presentation, find all of our new hires in the presentation and tell me how to pronounce their names.

You know, that's absolutely something a chief of staff would have done before, you know, and I and it was able to do it in 10 seconds, you know, and it's, it's pretty amazing.

So, uh, you know, uh, I would say over the past few months the the adoption of agentic tools is also going up.

No, it's not maybe scaled up the way people thought it would happen like, you know, but, but everybody is now figuring out a lot of new things, especially this week.

You're seeing very interesting products come out.

Um, so, so, uh, our enterprises, uh, getting the hang of it because, you know, initially they suddenly had the AI boom, then they had gen AI use cases, they had new use cases coming up.

Now, uh, agentic use cases are coming up.

So, so, so our enterprises getting the hang of it?

Um, you know, are they having to pivot their understanding of what to do with this, uh, you know, frequently?

So, so how are you looking at, uh, the entire scenario?

What is your advice maybe to enterprises on how to adopt it?

Yeah.

Um, so I think they, I think they are figuring it out, but I do think it's a process of trial and error.

There's, there's no methodology and I, I think that's okay.

Um, at this point, I think I, my, my advice we, we do this internally at Slack.

My advice to customers would be to try as many things as you like fiscally, responsibly can try um, and then see what works for you, your company and your individual users.

Um, we have tried to make Slackbot something that will work for all of your users.

And we think it will because it's close to where they're talking to their colleagues.

It's connected to your Slack and their system.

So it's gonna be very relatable.

It's it's easy to find very relatable.

Um, but we experiment with all kinds of other AI products.

And I think what Slackbot can ultimately help do is when you find the things that, um, are meaningful for your company and you deploy them into Slack so that people can use them, Slackbot will make referrals over to that and actually help people discover the AI that matters to you as a company.

And get them to use it, you know, and so I think that's the, that's the role that we can play.

But I would tell everybody, just go try as many things as you can, you know, rapidly.

I, I, I think our old, our old ways of thinking about capital expenditures, of building business cases like that stuff is completely invalid right now.

Yeah, more important that you try things as fast as you possibly can because the the rate of change is incredibly fast right now.

And if you're betting on business cases and long evaluations, like, you're just gonna take too long and you'll have missed the cycle, you know, so, yeah, so, um, also I know you spoke about, you know, agentic enterprise like, you know.

So so how so you mentioned Slackbot as being, you know, sort of like your personal assistant, but but how is the agentic AI piece helping enterprises, you know, take care of a lot of things, right?

Like, you know, you can you can build in a lot of, you know, um, as in regular functions being sort of hyped out to the agentic enterprise to take care of?

Is that how you would look at it?

And, um, and then your employees themselves working on really deep stuff?

Um, I'd, I'd say it's a, it's a combination.

So I'll talk a little bit about like what we've seen amongst our customer base and what we're doing ourselves within Slack.

So I think Slackbot gives you a basically a personal assistant that you could or chief of staff as you said to make available every employee.

And that's gonna be a productivity improvement for every employee.

I mean, it's gonna save them time and give them time to work on higher order tasks, higher, higher value tasks and, and be more creative.

Now the agents that we're, we're seeing thousands of agents that are being deployed into Slack and that's everything from, uh, coding agents, uh, to agents that are taking care of things that used to be handled through workflows.

Um to new agentic workflow systems.

Um, but I think the the probably the best example use case is actually a coding agent.

And so we've, we've got a number of those that have been deployed to Slack from Versal to Claude Code to open AI's Codex.

You can use all of those in in Slack now.

But if you think about um where the conversation is happening about what you want to build or about the changes you want to make to something you've already built.

Like it's typically happening in Slack channels, right?

Um, it's happening in team channels or an individual feature channel that where you're saying, hey, you know, this part of our own, our own devices website, we should actually update the copy through the logo over here.

Do that.

And people are talking about that in channels.

Well, these agents can just be another member of the channel now.

They can listen and you can say, hey, Cloud Code, just go ahead and look at, look above here and see what we talked about and go make that happen or codex, go do that, you know.

Or Versel go do that.

And I think that's, um, probably one of the first places many companies can start and achieve a ton of value is, is honestly from the coding agents.

But we're seeing it, you know, we're happening.

You see it happening in design and other areas as well.

So the way, uh, you know, the way a lot of companies have, have integrated Slack into their workflow.

So it's an integral part.

It's like your spine.

Uh, so, so is that a great place, uh, in a way for the agent to sort of rest because it gets the entire context of the organization because exactly what you said, the conversations are all there, right?

So it has the context.

It knows what is maybe taking the mind space of the company at that point, like, you know, so, so, so, so how are you looking at that?

So, you know, I hear a lot about orchestrating these different agents. But, um.