Slack Dev Day unveils Agent Kit and Add to Slack (Full Transcript)

[00:00:00] Speaker 1: It's important for WordSmith to be in Slack because that's where work happens.

[00:00:27] Speaker 2: Slack is the place where people are working, and you have multiplayer conversations.

[00:00:33] Speaker 3: That's where you work, and we don't want to take you out of where you work.

[00:00:36] Speaker 4: You interact with these agents with conversation, so Slack felt like the perfect solution.

[00:00:45] Speaker 2: The advantage of multiplayer vibe coding is that my whole team can contribute to the idea.

[00:00:51] Speaker 1: One of the really magic things about our integration into Slack is that in a context that's required to answer questions in the future is already in there.

[00:00:58] Speaker 3: There's so many different ways that you can build custom agents inside of Slack.

[00:01:06] Speaker 5: We don't have to build a new interface here. There's already one.

[00:01:08] Speaker 6: It's just called Slack. You just inherently understand it.

[00:01:11] Speaker 7: Now you've just got this extra person that lives in there and is able to pick up those tasks for you at the touch of a button. I think we're honestly just at the beginning, scratching the surface of what agents can do in Slack.

[00:01:23] Speaker 5: A future that we'd love to build is one where things are just happening in Slack.

[00:01:27] Speaker 8: Managing all the agents that are performing work, I couldn't imagine doing it from anywhere else but Slack. Hello, everyone.

[00:01:34] Speaker 5: Welcome to Dev Day.

[00:01:35] Speaker 9: My name is Ro Garcia, VP of engineering at Slack. It's so great to see so many familiar faces today and also some new ones. But especially I wanted to say hello to everybody watching us online today. Thank you for being with us here today. And today's a special day. It's a celebration of this community, developers, that has made Slack such a loved product. Thank you for joining us today. We have a packed 45 minutes with announcements, news, and also celebration of the work we have accomplished together. So without further ado, let's dive in. First, I want to start by saying thank you. Slack was built by developers for developers. And since the very beginning, back in 2013, you've been with us in this journey of exploring how an open platform can help to make working life simpler, more pleasant, and more productive. Thank you for being with us here today and being part of this community. But we know that there's a lot going on. AI is changing everything. And the opportunity for developers is massive. And we know that, for example, developers are writing multiple projects. And that's why GitHub announced recently that they have seen a growth of more than 98% year over year of AI projects. Also Stack Overflow announced in the latest developer survey that they have observed a growth of more than 84% in terms of developer adopting actively AI coding tools and also exploring to add it to the developer workflows. Finally, Price Waterhouse forecasted that AI could add more than $15 trillion to the economy by 2030. That's a lot. And also, pretty soon, just probably three or four years away from now. But also there's another side of the same coin, which is noise. Every week, there's a new model, a new SDK, a new interface, a new way to build agents. Also, product managers, and especially executives, and also customers, want agents and they want them yesterday. To our research, we know that developers spend up to five weeks a year managing this complexity and the constant ever evolving landscape of AI tooling that they have to manage. But it doesn't have to be like that. And I just want to make an exercise together here today. I bet that all of you pull your audio phones, you have Slack right now, and you are on Slack. And that makes total sense because your teams are on Slack. You are on Slack. You're making decisions. You're working in context. You're working in projects. And so, how about that applies to the AI era? And this is where our mission comes in. Because since 2013, our mission has been making working life simpler, more pleasant, and more productive. And that also applies to AI. Because we're on a journey to connect humans with agents in an experience that's delightful, human, simple, and perhaps even more quirky. Why not? And so, we want you to join us in this journey since we have been doing this for the last 13 years. And there's a lot of excitement ahead to continue making this true. So, what this means in practice? What it means to unlock AI for every employee? And there's three primitives, three silos that we have to connect. The first one, all your people. Your teammates, your customers, your collaborators. And also, your AI, AI assistants, AI agents. And finally, every platform. That means in practice, app, workflows, data, APIs that fundamentally define the tools and the data that your agents need. But we know that business and teams doesn't operate in silos, right? Because we need to work together. And that experience of AI needs to be multiplayer, which means in practice that we need to meet you in the flow of work, meeting you where you are, your teammates, your colleagues, and your teams. Perhaps now you start seeing where I'm going with this. And we start looking through these lenses, we start seeing that a work-operated system start emerging. One that allows you to ask anything and do anything. Well, I have news for you. Because Slack is that work-operated system that connects your people, your agents, and every platform you use to the power of an open ecosystem and a seamless experience with Salesforce. We have worked really hard together to make this a reality, and it's here. Because Slack is the only platform where AI is multiplayer. But we know that the companies building the AI future, which is happening now, are building for Slack. And that makes total sense. Because when you bring your agents to this flow of work I'm describing, something magical happens. When your agents can interact in a seamless experience with your teams and your channels and your canvases, unlocks a totally new set of values. But don't take my word for it. See what Kat from Anthropic recently said, that Anthropic runs on Slack, is the core operative system of our company. Context lives where work already happens. And I think especially the concept is very important. Context lives where work already happens. So what this means in practice, and we are so excited for the numbers that we started to observe for the last year, 300% growth in agents created weekly on Slack, 300% growth. More than 900% growth in AI users. We have also more than 1 million weekly active users in the Slack NCP server. But the numbers I'm the most proud of is that the number of more than 65,000 developers that have joined our developer program. And also the more than 125,000 AI apps and agents that have been deployed just in the last three months. That's a total new set of impact that we have never seen before, and all of you in this room and all of you watching us online are helping us to build together, and for which I'm really grateful for. But we know that for sure that how you build with AI has changed. Until not far along, when you started the journey of building agents, you start from idea to production, and this steepest curve of the creation cycle was so complicated. It would take quarters to build an agent. We've been really busy with the team at Slack to solve this problem, because precisely we know that your business needs answers and they need it now. And as soon as you can bring your agents, you can help your team to be successful. So we've been really hard working to create this experience that's fast, smooth, and low

[00:09:10] Speaker 5: friction.

[00:09:11] Speaker 9: So, I have some news to share with you all today. The first one is today we're releasing a Slack agent kit, which in practice means we have the bold SDK that all of you love and know through the years, and at the top of that, we have built a new developer experience that allows you to create agents on Slack with one-click deployment. That means in practice that your agents now can be in the flow of work in a Slack team meeting your team and your colleagues in the way you work. That's a new total unlock we haven't seen, and honestly, also tools finally match the ambition that we've been looking for. But we're not stopping only there. Today we're announcing the limited release of Add to Slack. No matter where you build your agents, in Brazil, in OpenAI, on your own stack, now you can deploy your agents to Slack with one-click. No previous knowledge of how Slack works for you to do this, and all your users can create these agents in all these platforms and deliver in real time for all your users. But this journey is fun. The journey is exciting, and things are moving fast. But also matters who also was with you in that journey. And I'm especially grateful for one partnership that has been since the very beginning of this new journey with us, which is the team at Vercel. Guillermo, Vercel team, thank you for believing in us in this journey to build this Add to Slack feature and continue our partnership to make this experience great for our customers and developers, and you all being since the very beginning working, of course, in a Slack channel and helping us to prototype the path to deliver this great experience. We're really grateful for it, and we cannot wait to keep working together. So that's a lot of words. Let's see it live and let's see it working. Now let's welcome Guillermo to do some demos.

[00:11:08] Speaker 10: Good morning and hello. I'm Jeremiah. I'm a staff developer advocate here at Slack. And as a developer advocate, I get to talk to a lot of developers. And earlier this year, I was at TDX, and I got to talk to one particularly disgruntled developer. And he was disgruntled not because his job got hard, he was disgruntled because he made the unfortunate decision to become important at work. You see, James had been at his company for 10 years, and he just gathered a bunch of knowledge in his head. And that means that everyone came to James. Can you help me get access to GitHub? Can you onboard the new hires? How do I fix this JIRA thing? And so James came to me at TDX, and he said, Jeremiah, how do I offload this to something else, someone else, like a bot or something? And James loved what I had to say next. I said, it's never been easier to build on Slack, to go from I have an idea to this is a working thing that people can use. And for most people, that journey starts right here. The Slack developer program. The Slack developer program has a lot of amazing features. I'm not going to cover all of them. But I will cover my favorite feature, which is sandboxes. Because since the dawn of time, if people wanted to create a Slack application, they had to spin up a free workspace, which was problematic. Or they had to develop directly in production, which was also problematic. And so we heard that feedback, and we created the developer program with sandboxes. With sandboxes, you can create up to ten at a time. They default to 180 days, and you have the option to extend if you're still working on something. And it's very easy to provision a sandbox. You click this button, and you fill out some basic information. My favorite option is here, because you can actually seed workspaces with real data, different data types, images, links, so that when you're building applications, it has real data and conversations and threads to work with. Now I've already built out and clicked those buttons and created Slack Dev Day. Slack Dev Day has these channels that were built. And now I have a working ecosystem, a working environment for James to bring his idea to life. But he still needs to build the idea. So this is a terminal for the vibe coders in the room. And we have the Slack CLI installed. So from here, I'm just gonna write Slack create agent. And I get to choose from a few options. And I think that for James, an IT support agent would be kind of what he's looking for. So we're gonna start there. We're gonna pick JavaScript, of course. And then we can choose our provider, if we want to start with Cloud or OpenAI. Let's choose Cloud. And we'll call this virtual James. Hopefully you all can relate. I wish I had a virtual me that could answer and do all the work that I don't want to do. So this is what we're gonna try to build. So what we're doing now is we have some repos that the Slack developer team has created. And we're essentially cloning it and kind of molding it into our own application. So I'm going to change directories into virtual James. And we're up and running. Before I hit Slack create, though, I need to connect my application to that workspace that I just created. So to do that, very simple again. I'm gonna type Slack login. I'm going to copy this, head back over to Slack, and paste it into any channel. Here we're going to approve this handful of permissions, take a second, look over it, and click confirm. And once again, we're gonna complete this handshake by copying the challenge code, heading back to the terminal, and pressing enter. Sweet. So now we have our development environment connected to our workspace. And now let's see what this app can do. So I'm gonna type Slack run to start up a local version of our IT agent, our virtual James. And we're gonna get it to the right environment, which is Slack Dev Day. So after a few seconds, our app should be up and running if live demos are working correctly. And they are today. Thank you. So I can come into channel, and my agent has already made its way to my environment. And immediately I'm able to interact with this virtual James. Now this is just a template. It's set up to handle most of the work that virtual James would need to do. But most people miss onboarding an agent. They get onboarded with Cloud or OpenAI, and they're like, ah, chat GPT is not working for me. But you need to spend just a few minutes onboarding the AI to know how you talk and the resources of truth that make your agent an actual virtual version of you. And with MCP and RTS, that has never been easier. It's just one link, but we'll see that in the next demo. But Slack was always designed to be the 2% of your IT spend that enables the other 98% to be more valuable. And that includes agents. So no matter where you're building an agent, you can automatically and instantaneously get it into Slack. So we built an agent in Brazil, and we wanted to get it into Slack. And with that new feature that we just launched, we were easily able to do that. So the feature's called, once again, Add to Slack, which just abstracts all of the details away from the developer. So they say, I have an agent, I want it in Slack, I'm gonna click this Add to Slack button. And in doing so, we'll handle the bot scopes, the permissions, getting it into your environment so that you can focus on the things that really matter, instead of the little tiny commas and details and errors. So that was a triage agent built in Brazil. And I can actually interact with it if I say, hey, I'm having problems with this blog on the homepage, can you help me out? Now this is a live demo, this is a real working agent, so it will take some time to go through that, but I've already had that conversation prior to going on stage for the sake of time. And it's really working. So I showed James all of this, and he was so happy, because he finally understood, if I have an idea, in maybe 10 to 15 minutes of time, I can make my ideas come to life in Slack. Back to you, Rod.

[00:17:35] Speaker 9: Thank you, Jeremiah. That was awesome. Thank you. And as we saw the demo of Vercel, now I would like to invite Char to talk a little bit about our partnership together and how it's been this journey. Thank you, Char, for joining us here today. Thanks for having me. Thank you. So, Char, in this partnership, we've been working together to build agents and connect the platforms. I'm curious, how Vercel thinks in the context of how developers are constantly switching, going back and forth, and meeting them in the flow work. I'm curious how you folks manage that, thinking about that.

[00:18:11] Speaker 11: Yeah. At Vercel, we use Slack a lot. Where the work happens, so we have product decisions, we talk about architecture, we troubleshoot problems, we make decisions, so there's a lot of context in there. And when it comes to bringing work into the collaboration layer, two things stand out. One is bringing our agents into Slack, so pretty much what Jeremiah showed in the demo. These agents are connected to third-party tools, so they bring the remaining portion of the context. So, instead of you switching between tabs, switching between the apps, finding or copying and pasting Slack threads into somewhere else, you just work in Slack. The second thing is working on the go. And I think it's now more critical than ever to work remotely on mobile, so one example was last week, we had an engineer who flagged a problem on the dashboard, and we went back and forth on the thread, I tagged V0 on that thread, and V0 caught a PR, merged it in less than 15 minutes for something that could have been until I go back to the laptop or the next day. So, it makes work much faster, and it keeps everything in one collaboration layer.

[00:19:29] Speaker 9: Amazing. So, sharing context and also being able to work in the go. Taking it from there, I'm curious, what have you observed in your experience with your customers and developers, what are the patterns of building agents that you say is working extremely well right now?

[00:19:47] Speaker 11: Yeah, so Vercel is an agentic infrastructure, so we have all the primitives and abstraction layers for you to build production-ready agents, and what we have seen both internally and externally with users, we ship a lot of agents, and internally, some examples I can give you is our D0 agent, which is our data science agent, it's built on Vercel, and it basically has access to our data warehouse, it understands our semantic layers, and it's exposed on Slack, so everyone internally can ask data questions from it. It answers more than 20,000 questions per month, so it takes a lot of load off of our data science team, and other examples of billing agents, which kind of acts as a pair programmer with our on-call engineer, you tag it in a thread, and it can actually look at our billing system, our invoicing system, and it can troubleshoot problems with that on-call engineer, which makes the job much faster, and we see the same thing externally with our customers, they're shipping agents on Vercel, and the ones that are new to it, they're very excited from going from zero to one very fast, and the ones that are actually building agents, they want to manage and build agents in one platform, and govern them in the same place, so that's been working really well.

[00:21:08] Speaker 9: Wow, and I love the use case, especially the on-call use case, and that shows that having defined pretty clear use cases help customers also to realize very quickly the value, and finally, I'm curious, then, we have talked a lot about how AI is changing the landscape for developers, I'm curious, what do you think are the skills that will define perhaps some of the success of the engineering teams in the future?

[00:21:33] Speaker 11: Yeah, I think my biggest feedback to engineers is now that software is becoming so much easier to ship, it's so easy to build anything, so having an understanding of how users actually build the thing, use the thing you built, and understanding what to build next is the most critical skill set, so I would encourage engineers to kind of span out to more product thinking.

[00:21:59] Speaker 9: Product, amazing. Well, Char, once again, thank you for the partnership with the Vercel team, you all have been incredible partners in this journey, and we will continue working together. Thank you. Thank you, Char. Thank you for coming today.

[00:22:13] Speaker 5: Thank you.

[00:22:14] Speaker 9: So, we've been covering the keynote, the topics that's driving this AI revolution, Jeremiah showed us a great demo that's introducing us to AgentKit, and now to Slack, we met Char, and talking about our experience, so now it's time to welcome Katie, that's going to walk us through what it means to be a trusted platform with AI.

[00:22:33] Speaker 12: So, you just heard how you can spin up and deploy an agent in just a matter of days, but really, we have to ask ourselves, what makes that agent actually good, right? What makes it work? What makes it effective for the people who actually have to interact with it? And it's so easy to assume that that's really about the feature set, but actually, what really, really matters is the user experience. Sure, functionality is, of course, important, I'm a product person, I live and breathe it, but what matters now with your agent is, of course, the context that it has, the way that it builds trust, and of course, the way that it really works. Now, it can feel like those building blocks have changed, but actually, Slack was designed for this precise moment. Brad might have showed you this a little bit earlier, this is our north star here at Slack. We make the working lives of people simpler, more pleasant, and more productive, and as a product person, I oversee our productivity surfaces, that's Canvas, huddles, lists, all of the interop work, and every time we make a product decision, we come back to this. And when I think about building agents, it really follows the same principles. First, the experience should be simpler, it should help take the friction out of those existing workflows that the agent is engaging with. It should be more pleasant, the agent should feel like a natural extension of your team, and ultimately, we should all be more productive. The agent should take the heavy lift out of the work, we should be able to get more done much, much faster. Now, when you take these core product principles, and you apply them to building on Slack, you actually sort of start with an outsized advantage, because Slack, and the context that Slack has about your organization, is something that no model can replicate. All of the decisions, all of the projects, all of the conversations about that work, that evolves every day, in real time, in the space that your teams are working. So when you deploy an agent to Slack, you don't just have an agent that can answer questions, you have an agent that can understand the who, the what, the why, behind the request. And then, as it promotes decisioning, as it promotes the next set of actions, that all happens in the context of where teams are working. It's not, as Rod said earlier, and actually, as you heard in the video to start, it's all about that multiplayer experience. We're not prompting you to some separate UI where you have a one-on-one conversation with an agent. That is core to what makes Slack the best platform for working with agents. So a few principles that we keep in mind when building a great agent. First, we just talked about it, it's that organizational context. The agent really needs to understand that who, what, why behind any request that comes in. Second, though, is human control. A human needs to be able to interact with the agent and guide and steer the interaction throughout the task. Third is progressive trust. The agent needs to earn more responsibility over time. And lastly is transparency. The agent needs to show its work so that you're never left wondering what the heck just happened. And this is all available to you out of the box with Slack today for whatever model you choose. Gemini, Claude, GPT, Slack was designed to be model agnostic because we believe that the difference is not the model, the difference is Slack. So what does this mean practically for you as developers, as builders? What tools do you have? Well, the first is Slack's MCP server. MCP is what allows you to actually tap into that institutional knowledge. It can query your channel history, you can pull threads, you can read files, it can really allow you to understand the full context of all of the conversations that have been happening in your Slack to date. In just a minute, you're going to see Saurabh come up and show a demo of an agent that's asked a question. But it doesn't just respond. It understands the situation. And MCP is that bridge between the agent and the institutional knowledge that makes the agent actually feel like a natural extension of the team. Second is BlockKit. All new richer, more interactive components, whether that's cards, alerts, carousels, new data visualization surfaces, and ways to really allow your agent's output to look and feel more dynamic, native, and polished without you having to do all of that heavy lifting and hand crafting yourself. BlockKit also provides buttons and dialogues so that your humans can interact with the agent and have conversations that effectively guide the conversation with that human in the loop the whole time. Work objects allow agents to present structured, trackable data in channel. They provide a transparent view into data that the current state of data that lives in third party sources. So it really builds trust and transparency in the experience. And lastly, of course, the real time search API. It grounds every response in what's happening in your Slack today. And every result is obviously permissions aware. Now these are all, maybe they feel like incremental improvements, but they're not. These are primitives of what makes a fundamentally differentiated agent experience. Now, enough of me talking. I think we're pretty much done with slides. It's my privilege to invite up Saurabh Sarni, our principal software engineer, who's gonna walk us through a demo. Let's see it. Cheers.

[00:28:48] Speaker 5: All right.

[00:28:49] Speaker 13: Thanks, Katie. You walked us through the primitives that make a good agent. Now let me show you what that looks like in Slack. Here's James. James has joined Pronto this morning. He got his laptop, access to Slack, and his manager just sent him an onboarding doc. He's going through it. He needs to get GitHub access. And he figures to get GitHub access, he needs to use IT agent. So he opens IT agent right in Slack. Basically the agent surfaces suggested prompts based on the context he's viewing in Slack. That's powered by Slack's events API dispatching assistant context events to the agent, and then agent responding back with set suggested prompts API, setting these prompts dynamically in real time. Now James clicks request GitHub access, and agent starts working on it. What you see here is agent sends back a plan block, clearly showing what it is doing, what it has done, and what's next. This is transparency. Agents can render everything step by step, all their reasoning, right in Slack. Now agent comes back with a confirmation message and shows James everything it wants to do before proceeding. Block kit makes it easy to structure these approvals and make them actionable. The agent doesn't act until James says yes. That is human control. And it's how agents build trust over time. James clicks confirm. Agent starts streaming tool responses. And all the provisioning is done right in Slack. No need to create a ticket. No waiting. Everything done in just a few seconds. Now let's switch surfaces and see what happens when the same agent shows up in a channel where the team is already working. A few days later, James is running into VPN issues. He goes to help IT channel and types, I'm getting kicked off VPN every ten minutes. Anything I'm missing? Now here's the magic. The agent, the same IT agent that you can DM, that you can add mention in channel, is also listening to every single message in help IT channel. And it jumps in whenever it can be helpful. But the real power comes from context. The IT agent uses Slack MCP server that taps into RTS, realtime search API, and searches messages across Slack. It reads channels, pulls prior threads, and reads files. And that's how it does this. Without this context, the best agent could have done is just provide some generic VPN troubleshooting advice. Now what you see here is industry standard markdown rendered by the agent. You don't need to handcraft blocks for typical LLM responses. Now Slack API accepts standard markdown that you can directly send to chat.post message. Developers have been asking for it, and we are so glad to bring that to all of you. Finally, the most unique part of this experience is this answer isn't just for James. Dave can confirm it, Priya can correct it, and another new hire can find it already answered. This is agents in Slack working with the team, building shared context for everyone. A multiplayer experience in Slack. That's the build. RTS API, MCP server, agent UX, and block kit. All of them working together in DMs and in channels inside Slack. Now the next question is, how do you get your users to use the agents that you have built? Nate is going to walk you through that. Thank you.

[00:33:25] Speaker 14: My name is Nate. I'm on the product team here at Slack. Our team works on the developer experience, block kit, the MCP server, the APIs that enable a lot of these integrations, the Slack marketplace that allows you all to distribute your apps to customers in Slack. We are so excited to partner with you. I want to talk a little bit more about adoption and getting agents that you build actually adopted in Slack. So first, after you build an application or an integration or an agent, does anyone ever feel like this cat here when you're trying to get people to actually use it? Right? You build it. You test it. You iterate on it. You polish it. And then the journey begins to actually get people to use it. The hardest part oftentimes isn't actually building the features or the capabilities. It's all the messy human parts of actually the change management, discovery, and actually getting habitual repeat use. And at Slack, we aspire to make this 100 times easier for all of you. There are already millions of teams working in Slack every single week. And many of them are having, you know, conversations and working with apps and agents already. And when you bring an agent to Slack, it is not a separate desktop app for users to install or a separate browser tab for them to bookmark. You're bringing the agent to where the teams are already working and having conversations. And most importantly, when you bring agents into channels, there's something really special that happens. There is a social element of discovery and awareness and learning that happens. People learn from how their teammates are using the agents. The prompts that are working and that aren't working. And they can help steer the agents together. And when agents come into channels, agents are no longer a personal productivity tool. They become a teammate that enables the entire team and the organization to be more productive. And the entire team and the agents to be more aligned. You can see what the agent is working on. You can observe it. Other team members can jump in and help steer its output towards what the team's goals are. So the entire team and the agents are working towards a common goal. And we want to take this a step further today. We are so excited to announce Slackbot as an MCP client. Yes? So, whether you're building a full-fledged agent or you just want to bring your MCP server into Slack, it has never been easier to do that today. We are so excited to partner with these amazing partners here on a pilot. This will become generally available this summer. And for developers, we want to make this process extremely easy. You can register your MCP server with your existing Slack apps. And when customers install your app, they immediately get access to the MCP server and the tools exposed through it. And can immediately start using it and connecting to it through Slackbot. So we are so, so excited about this. And really, we are just so, so humbled to get to work with all of you. We think it is the most innovative ecosystem in the world. We appreciate everything you do to partner with us to build out these great experiences. Over 2,600 apps in the marketplace. Over 2,700 customers and partners that have built on top of the MCP server already. And it is really such a privilege to get to build these experiences together with you for all of our mutual customers. And speaking of customers, InGen is one of our customers that has really started to put these agents and integrations to use. They have hundreds of tools that they work with every day. And they really see Slack as that central operating system to bring it all together. And so, I want to bring up Lauren to help demo this experience and see what it all looks like together. Thank you.

[00:37:59] Speaker 5: Awesome. Good morning.

[00:38:01] Speaker 6: Well, my name is Lauren Nielsen. I am Slack's technical product marketing manager here. And I think I might have, along with Jeremiah, the fun jobs for today where we actually get to showcase really cool functionality and product. And bring it to life. And again, what we're doing here is we're bringing these really cool use cases to your users. Awesome. That looks good. Fabulous. Well, I am, let's pretend I'm an account manager here at InGen, tying up a couple loose ends before my customer, Acme, has their massive sales kickoff in Vegas. And by the way, I always like to talk about the fact that I don't know why we do sales kickoffs in Vegas. I feel like that is the most unbelievably dangerous thing on the planet. Not my circus, not my monkeys. We are just here to facilitate. And I'm going to start this whole process to get those last little bits done in where I always start my day and finish my day too. And that's going to be Slack, my centralized hub. And specifically, we're going to take a look at our Acme company kickoff channel here. But we're also going to tap in our Slackbot AI teammates. And actually, this is where we're going to start. I'm going to simply ask Slackbot, hey, is Acme ready to go for their sales kickoff on Monday? And what the cool thing about Slackbot is, is that it has access to all of that context living in Slack. I've got my channels, I've got my canvases, I've got my files, I've got my, you know, my conversations. And what's going to happen here is that we're going to see this start to query this information from me, pull this context together, and it tells me, yeah, we're almost there. We're close. But there are blockers. But when I review this information here, we're going to notice one thing. The issue is that these are outside of the Slack ecosystem, right? These are in external systems. Specifically, in this case, we're looking at scattered across engine stack of we've got some in linear, right? And then we also have a couple different contracts that are going to be stuck in DocuSign. Okay. That's great. We know that information. But I can't action on that just yet, right? And that's literally why we built the Slack marketplace, because essentially what this is doing is giving us the opportunity to pull all the context, all the knowledge, all those different tools that you're leveraging on a day-to-day basis, up to 6,000 I think is what we have in our ecosystem right now, so that you can take these agents, you can take these apps that you're leveraging on a day-to-day basis and use it where you're already working, which is Slack. So, in this case, we're going to take a look at how we leverage linear, cursor, DocuSign, all these fantastic pieces. So we're going to go back to Slack now, and we're going to see what happens when we actually have Slack bot connected to that ecosystem. Same question, right? We're going to ask, hey, is Acme ready for their sales kickoff in Vegas next Monday?

[00:41:02] Speaker 12: And what's going to happen here is we're going to pull context through, coming in

[00:41:06] Speaker 6: from linear. And again, we're talking about permissions, right? This is really important. It knows that I have access to this information. It's going to automatically pull that context. And then it's also going to take a look at querying the different issues that are associated to the event. It knows what's going on, the context. And then on top of that, it's also going to cross-reference it with all that contract data that's living in DocuSign. So what I get here is not only just a summary like we saw before. That's great. We love that. But an actual ability to review this context in a work object where I am already working, which is Slack. Phenomenal. Let's see if I can close that. There we go. Perfect. Amazing. We feel really good about this. I understand what's going on. And one thing I always like to keep in mind is what we used to do before, right? This used to be going in a linear, reviewing all this context. Okay. What do I need to understand? You know? Then I go into this next contracting system. And really what happens is that you have to hold all of that information in your head to be able to get an end result. And I have a running joke in my household that I share a brain cell with my dog, and I am unable to keep anything within this more than, you know, what's necessary. Which means that I'm not going to be able to do that. And I know that. I know my ability. And we're going to keep it to utilize Slack bot. Cool. So we're going to go ahead and kick this off to essentially start these two pieces. Again, we have our DocuSign blocker, and we have our bug in linear. But what's great is that Slack bot is going to prompt me immediately and say, yeah, do you want me to create a channel, right? Let's say I'm not necessarily the most technical individual. I'm an account manager here at Engine. I need to be able to connect into those individuals that actually know how to do it. You guys, right? You guys. How do we fix this issue before we actually get on site on Monday? And what's going to happen is Slack bot is going to pull our individuals together. That could be the individual who is in charge of the linear issue. We have our lead engineers, we have our QA leads. But then there's also this one extra thing, and that's our cursor agent. So not only are we bringing our people together, but we're bringing our apps and agents together to work together in one single area. So when I open this up, we're going to see not only do I have all the context, but I'm also going to see the conversation is already flowing, right? Between Adam, our lead engineer, and the cursor agent. And the way that I think about this, right, is this is a phenomenal example of agent and human interaction, right? And I also think that when you think about Slack, we're in the business of humans. We're not trying to take away roles. We're trying to give you the ability to leverage the tools, the systems, the data, the information that you need at your fingertips in one centralized place for you to do your job better, quicker, and at a more sustainable rate. Awesome. So when we take a look at this, we can actually see that our cursor agent has already read the code, context, and it's already starting to suggest some fixes literally immediately, which is fantastic. And all of a sudden, we have a path forward here, feeling really good. And when we're thinking about it from the perspective of engine scale, this could really be, you know, this compression of coordination is the difference between shipping and slipping. Really important stuff. All right. So we have conquered one blocker. We now have one more. And that is to make sure that we have all of those contracts sent, delivered, signed before the day on Monday. And again, we're going to leverage our agentic teammate here by asking, hey, can you go ahead and pull those unsigned contracts and send them to the Acme procurement team? Now, what this does is similar to what we saw with Linear, I can pull all that context and knowledge. It knows that I have access to this information. It understands that I need to get the specific right templates. It's going to filter on signature status, so I'm not just looking at a bunch of different, you know, information. And what it's going to deliver to me is a fully built-out table for me to understand, cool, we've got four that we need to get done. Let's get them out of the way. And instead of just leaving it there, it gives me the opportunity to automatically say, yep, go ahead and send them now. So not only has it brought that information to me, it's given me the ability to action on my behalf.

[00:45:59] Speaker 5: Awesome.

[00:46:00] Speaker 6: We'll give it a moment to think and do this. But really, what we're doing here is we've seen contracts out, right? We've seen our bug has been fixed. But we're Slack, right? We're Slack. And that's also a Salesforce company, which means that we have all of this information that's living in a non-silent area. We have all of this context. So I think we're going to finish this off with a little bit of our nod to our Salesforce mother, I like to call it the mothership, our lovely mothership. And what we're going to do is essentially ask Slackbot to update the opportunity for me. And not only that, I want you to go ahead and send a note within my channel to let everybody know, awesome, we're good, and we're feeling strong about this moving forward on Monday. So not only can we see this, it's going to pull this message, it's going to send the details, and it's also going to update that opportunity if I even pull up the dropdowns. Accorded Salesforce sent message, done. That's how easy it is, right? I don't have to go into my Slack, or my Slack, my Salesforce ecosystem, I don't have to go and hunt down a record to be able to understand, hey, what is it that we need to be able to do here to make this work? The intent is that it's all living within your Salesforce and Slack ecosystem, which is fantastic. Now, we're doing a lot of this here within engine, but in reality, we're already doing this within Slack. And so what I would love to do is for the 19th time, bring up Rod, and I'm also going to bring up Shravini, and we're going to go ahead and have more conversations here, fantastic.

[00:47:35] Speaker 9: Awesome. Thank you so much. Oh, my goodness. A lot of contents, a lot of demos, and a lot of announcements, but there's another style also this revolution is happening, which is also challenging us as developer community to rethink about our roles. So we wanted to take our time a little bit to talk about that, and that's why we have Shivani. Shivani, thank you for being with us today. I know you really well, but our community would like to know a little bit more about you. So can you tell us a little bit about your role at Slack?

[00:48:06] Speaker 15: Thank you, Rod. Hello, everyone. I'm on the developer productivity team at Slack. So our job is to make engineers like developers more productive. So my day-to-day job involves like working across different code bases, a lot of context switching.

[00:48:20] Speaker 9: Thank you. Thank you. Shivani, I'm curious how AI is helping you to rethink your role, and how has that been evolving in your experience?

[00:48:31] Speaker 15: So at Slack, we use a few AI tools, including custom agents within Slack, which helps engineers with escalations, incident bots. So for me, I personally use like Slack bot quite a bit. It helps me reduce like context switching. So I can like stay in the flow while I'm where I work. So Slack bot, internal Slack bot also has integrations to our internal MCP, so I don't have to break out of my conversations to go search or do actions. So I basically get updated on the incident channels, my project work, my JIRAs, and my meeting schedules, everything from Slack bot, like right where I work. But there's also times, you know, where I want to do individual focused work, prototyping, so I write coding. So on my local host, like I have also integrations into different MCPs, including Slack MCP, to bring context across different data sources, right into like where I work. So I think overall, I would say like, about over 40% of my engineering time is like, it's saved like with all usage of all of these AI tools.

[00:49:43] Speaker 9: Thank you, Shivani. Context, tooling, and keeping in the flow work. I'm curious, Shivani has done an excellent job in Slack together with the entire developer experience team internally and running out the AI tools for the team. Shivani, in your experience in this rollout we have done for the entire engineering organization, what has worked well and what building blocks developers are using now to develop amazing

[00:50:06] Speaker 15: products? So at Slack, like Slack engineering and non-engineering teams, like across Slack, over 99% of Slack teams are using AI tooling now, and we have seen over 40% of PR throughput increase. So some of the learnings, this didn't come like, you know, top-down mandate. It happened because we focused on two key accelerators. One is show and tell. So we hosted these weekly meetings and dedicated channels where engineers would come and share how AI is helping them, you know, automate or ease some of their workflows to enable other engineers. And reduce the onboarding friction. So we have automation to help set up, like, all the AI toolings and the skills, like the base skills that are needed for engineers to be effective from day one. So we rolled out all of those. So while doing so, we also hit some friction points. So most of the AI tools right now are intended for single-player use case. So engineers are, you know, working in silos. And the other thing is, like, abundance of tooling, right? So there's so many AI tools. So it has caused, like, fatigue for engineers in exploring and trying out all these different engineering, these AI tools. So our team helped, like, evaluate and roll out the tools and the skills that are most effective for our team, for our Slack and for our environment. Like, so, yeah, that's how the rollout happened.

[00:51:32] Speaker 9: Amazing, Srivani. Thank you so much for everything you do. And also thank you to all our team in Intent on Engineering for helping us to navigate this wave. Thank you, Srivani.

[00:51:41] Speaker 5: Thank you. Thank you.

[00:51:43] Speaker 9: Thank you. Thank you. So let's summarize what we have seen for the last roughly 40 minutes. Context, multiplayer, and ecosystem. The three pillars of the Slack advantage that we've been seeing today through multiple demos and conversations. But also I wanted to share a shout out, again, to our community. And we have our Slack developer, a Bowsy board, which is a board of developers that help us to shape the future of our products, helping us with, you know, how the developer experience and feeling what's happening in the ecosystem and helping us to improve and make Slack continue being such a loved product. But there is one of the world members that's very special to us. Scott, I want to invite you to the stage. Please come in. So while Scott gets to the stage, we started the journey of Platform 2.0 internally many years ago. And Scott has been by the very beginning with us sharing his experience and your own, you know, what you've seen in your company. So Scott, can you tell us about the moment you fell in love with Slack?

[00:52:55] Speaker 16: Sure. Our organization started using Slack about six years ago. And we moved over from a different product. And that was about the time that Workflow Builder came into Slack. And so I just fell in love with Workflow Builder and all the automations that it could provide. I started using Workflow Builder to automate a lot of things within Slack. And I did some amazing and great things with it. But then I also saw the limitations of it. And so I thought, why not build some custom steps for Workflow Builder? So I used the Bolt SDKs, built some custom steps, and created a bunch of amazing and things that Slack didn't even know Workflow Builder could do. So that started my journey with Slack development. And I just fell more in love with the product.

[00:53:45] Speaker 9: Thank you for falling in love with us. This is awesome. And, you know, in the Slack channel we have with the developer, I will see Scott sharing feedback, sharing his customer steps and say, oh, my God, we need to meet Scott and chat more often. Scott, what makes your experience different from, you know, building in a Slack from other platforms?

[00:54:05] Speaker 16: Sure. The experience with Slack, I didn't originally start off as a developer. I was just a regular IT support engineer. So the interfaces with Slack, Slack gives you all the tools to build with Slack. Everything from the documentation to the Bolt SDKs to new SDKs, even the GitHub repositories that Jeremiah showed and the samples, Slack gives you everything that you need to build on the platform within that platform, and they make it fun. Amazing.

[00:54:41] Speaker 9: I love it. And now we're entering to the genetic era, Scott. What's your feelings? What are your thoughts about everything you have learned about building agents on Slack?

[00:54:52] Speaker 16: Again, Slack is where work happens. And they want they have this concept of being the whole agentic OS for everything. So Slack has provided us with a lot of different tools to build on the platform. Everything from the new assistant classes. We saw new blocks in the block kit builder, MCP servers, RTS. Again, they're giving you all the tools that you need to build on the platform and keep everything all productive and efficient within Slack.

[00:55:26] Speaker 9: Love that. So let's do a real talk. Of course, there's a lot of good things, but also things to improve. So Scott, tell us one thing. One thing that you think you would like us to improve. This is the moment. You can do it. Tell us one thing you would like us to improve.

[00:55:42] Speaker 16: What would I like you to improve? Well, two things come to mind. Do we have time? Please. Yes. Do it. First thing. If Slack is going to release a feature, release the APIs for it, please. This allows the developers to not only learn the product, but to help our customers and our users learn the product as well. So by releasing the APIs, again, it just involves that whole ecosystem of using that new feature. And in addition, with the developer advisory board, listen to the developers. We're telling you what we need, what we want, you know, fix some longstanding bugs. I'm not saying there's many. Fix longstanding bugs. Listen to the APIs that we want adjusted so that we can help you and everybody use Slack and, again, become more productive and efficient.

[00:56:39] Speaker 9: Love that. That wasn't that spicy. Yeah. Give it to Scott. The good news is that we have a lot of product managers here. So great feedback. Amazing. Down here. So, Scott, thank you for being a great partner, a customer, and helping us to shape our community. Thank you for being here today, Scott. Thank you. So three ways to get hands-on experience today. First, join the Slack hackathon that we're starting today. So you can pull out your phone, see all the QR codes. Second, join the developer program. That's the best way for you to stay at the top of everything happening in the Slack platform. And finally, help us to shape our community to continue improving our products and platform. So with that, thank you, folks. This is the keynote. We'll see you in a few minutes for the product partner.