Name: Lucas Isaza, Block
Uploaded: 2026-11-11T00:00:00.000Z

More from Block Builderlab Launch Event

Lucas Isaza

Product Lead, Builderlab

Block

19:35

18:15

08:26

Info

Lucas Isaza, Block

Lucas Isaza

Product Lead, Builderlab Block

(INTRO)

Alison Kosik

>> Welcome to Block Builderlab Launch Day 2026. I'm Alison Kosik and I've got Paul Gillin. He's the enterprise editor with SiliconANGLE. I've got him right beside me. I want to bring in our first guest who's joining us on theCUBE. We've got Lucas Isaza. He's the product lead of Buzz. Welcome to theCUBE.

>> Thank you for having me.

Alison Kosik

>> It's an exciting day, isn't it?

>> Yes, very exciting.

Alison Kosik

>> It's a long time coming, I'm sure.

>> Yeah, it's a combination of a lot of things that we've been building and working on and using internally. And now today is the day that we get to finally share that with a bunch of other customers and folks interested.

Alison Kosik

>> For someone who's never heard of Buzz, talk me through what is it and how is it different from the AI tools that most of us already use and know about?

>> Sure. So I want to maybe take a step back to most of our work is collaborative and whether it's coordinating an event like this or launching a product, And a lot of the tools that we use to do that work are single-player, meaning I use that tool and then I take the output of that tool and share it with my team. And I think our realization, or what led to Buzz, is that AI makes us really powerful at those single-player tools, but we become the bottleneck to doing more as a team because we need to then take that information and share with others, or vice versa. And so the idea behind Buzz is It's a collaborative workspace where both humans and agents can participate. Agents have their own identities, so they look like humans. You can tag them like you would a human. And it just makes working with our team much more effective because now you're able to tag in the agent when you need to go off and do an investigation or write code or anything like that. And that's really the premise behind Buzz. There's a lot of other pieces, like it being open source, that really matter to us as a company that really cares about sovereignty and giving customers the ability to run it on their own infrastructure if they want. But at the highest level, Buzz is a collaborative workspace for both humans and agents to work together.

Paul Gillin

>> You really have to see the demo to understand, to appreciate what this does, because I think we understand Slack, we understand different workgroup tools that we use, but what you've done is integrate agents almost as people in the workflow, where you're erasing the distinction between agents and people. Was that one of the objectives?

>> That's right. The idea was to, I think, make agents have more control, more things that they can do on behalf of their human owner. And what we've done with Buzz is really give agents their own identity. And like Brad mentioned, every action they do is auditable. And so I think really bringing agents as real participants in the work that we're doing is what we're trying to do here. And then maybe the broader Builderlab, which we didn't really mention, the Builderlab is the broader set of ecosystem that makes Buzz really useful for us. So it's everything from the tools and authentication to managing your team and being able to set up remote agents, which

Paul Gillin

>> I think the audience is important here. Obviously, today your audience was clearly developers and they loved it. Are you aiming at an end-user audience as well, or is this strictly for people who build things?

>> Yeah, we're building these tools to be the best tools for us to use at Block as employees, and that spans all functions from the product manager to engineer to designer, all the way to the support specialist, the sales rep, and the account manager. So we're building these tools to work best for us internally. What we've learned over the last couple of months since we've been talking to external customers is that there's a really wide set of users that benefit from Buzz. And so we're not making this product— we're not making it work perfectly for someone in the oil and gas space or something like that. But what we found in talking to customers is that there are some oil and gas space related companies that are using Buzz. And I think a big part of that is just in how customizable it is with the plugins that Brad was talking about. because it's such a malleable interface, we're seeing that it actually ends up fitting a lot of other industries and customers that we wouldn't have expected.

Alison Kosik

>> —can you take me through a typical workday using the tools that we're talking about?

>> Sure. So I think we're using— so we use Buzz to collaborate as a team, whether it's investigating an issue. So we can go through that example. We find a bug. And the thread there is around, hey, this isn't working correctly. What should we do to fix it? Or as we were getting ready for this event, we still need to add this feature or this capability and it's not working. And that thread might be a discussion around the trade-offs, which way should we take this? And let's say we decide on, hey, let's take approach A, then someone else tags their agent to do approach A and finally you have the code change. And so that's when it comes to making changes related to the code, but I think On the other hand, I'm handling a lot of the support emails that we get for Buzz. And so one unique way in which I'm handling that is I have an automation that runs every day at noon. It scans all of the emails that we received to help@buzz.xyz. And then it's helping me triage. So it's helping me find which of these 50 emails that we received are emails from customers that need to delete a community or that are having trouble setting up. And that really helps me know who I need, how do I prioritize my responses and my time, and also what we need to escalate if we need to escalate. And so I think there's a lot of the use cases that we talked about today was for engineering-related work, but one way that I think it extends to everyone else is with Builderlab, we have this apps platform that we just talked about. And the apps platform really allows anyone at the company to build a website and share it with anyone else in the company. And surprisingly, the biggest adopters of the apps platform have been non-technical roles, meaning the account manager can now build their personalized dashboard that pulls data from their email and Salesforce and Google Sheets and whatever other data sources. And so everybody is able to now build their own dashboard. They're not introducing a new source of truth, so it's relying on the same sources of truth that we already have. And it's a really safe way to share it too, because that account manager can share the link with me, and the app is going to use my credentials to view that dashboard. So if it's pulling from his or her Gmail, it'll be pulling from my Gmail if I look at it. And so the Apps Platform really allows us to have a new interface for doing our work. So a lot of it might be collaborative and talking in Buzz. But a lot of it, too, might be trying to understand trends or trying to understand, interpret feedback that we're receiving. And a lot of that I do through apps that I build. That help me digest all of the information we're receiving from different sources.

Paul Gillin

>> The architecture is customizable, it's plug-in. The plug-in aspect is a really interesting component of it. What was your thinking in designing the agents to be so configurable and extensible?

>> Yeah, so a big part of that is the evolution that Alex and John shared of agents having a persona in some ways, having an image, something that makes them recognizable, something that makes them fun to work with. But the whole idea behind plugins is that we initially had the Buzz, the first version of Buzz that we launched in July. And there were no plugins. And what we realized is we had all of these requests or ideas from users in Block and externally saying, "Hey, I want to add this," or "I want to add that." And we realized that we were the bottleneck to enabling that. And if you want to have a plugin that allows you to see your work in the best way possible, you could only do so if we were the ones to review, approve it, and incorporate it. And the whole idea behind plugins was let's remove ourselves from being the bottleneck and actually design the app with an architecture that from the very bare bones supports customizability, meaning anybody can now build a plugin and that plugin can represent anything on the app. So that's really been the journey towards getting to plugins. And since launching, what we've found is there are so many interesting ways to use plugins or to build plugins that we're now similar to apps, seeing folks build their own and share them. And it's kind of becoming the best account manager plugin is pulling data from all of these sources and helping you figure out or for all these other use cases.

Alison Kosik

>> Have you found that you've been able to perfect Buzz just by watching how people interact with AI agents and maybe seeing their frustrations or seeing the shortcomings?

>> There's a lot of that. That's definitely been a really important source, both at Block and externally hearing from others. I think the one big piece is that we have all of these users now using Buzz and sharing feedback. And what we're really focused on is managing a user's attention. In a world where you have multiple agents and in a world where you can do so much more, the hardest part is figuring out where to spend your time. And even in Buzz today, many of us have multiple channels where we're doing multiple things at the same time. And it gets hard to know what should I focus on next. And so that's definitely, I think, a point of frustration or something that we're really working hard at. Trying to make easier is how do we make it easy for you to know what to focus on in this world where there's a million things going on. I think the other piece, too, is we have this channel— and this kind of goes to how we're addressing things natively in Buzz. We have this Buzz Bugs channel. And anyone is posting issues they see with the app there. And we've set up an agent. That listens for messages that look like a bug using what Brad just mentioned. So you can kind of— I set up my agent to track all of the messages posted, and if something looks like a bug, to reply. And so my agent is creating the ticket for that issue. maybe this button doesn't work or that thing doesn't work. It's creating that issue, and then we have a bunch of the team with the goal of take on one bug a day, you or your agent. And so teams are pushing their agents to help tackle these issues. And we're starting to work towards— what we want to get to is more of a machine where the agent creates the issues and we have other agents that pick up those issues. But there's certainly a ton of really, really good feedback that we're hearing from the folks at Block and XDR. Currently right now.

Paul Gillin

>> One thing you said in your remarks was that you had to change the mindset toward getting away from the chat interface. What did you mean by that, and how does Buzz move away from the chat interface?

>> Yeah, so I think the— what we want to move towards is a world where the work that you're doing, you're doing it on the interface built for that in the way that you like it. And if that means you're better off understanding a set of numbers through an app that helps you visualize that information, we want to make sure that the product can push the user to that interface for that experience. For the same reason, plugins can be fully customizable. And so what we want to do with that is we want to make sure that you're not constrained by the interface. Your work, the best way to do it is through a plugin, we want to allow you to build that plugin. And I think chat is still important. I think maybe the problem or the thing that we need to figure out that we have a lot of experiments we're working on is how do you manage someone's attention when there's 5 chat sessions going on at the same time and having a really clean way, a design-centric way towards making that approachable for the best engineers, but everyone from the best engineers to the best sales reps to the best account managers at Block and hopefully beyond Block. It helps others too.

Paul Gillin

>> So you're also open sourcing the technology. I know that's an important cultural dimension at Block, but commercially, what is the thinking behind open sourcing Buzz?

>> Yeah, certainly. So today we launched Builderlab, and Buzz is a really big part of that. The thinking behind open sourcing is we want to always give anybody using Buzz the option to run it on their own infrastructure. And so by open sourcing Buzz, we make the code available for anybody to download and change and use and run on their own infrastructure. But we also realize that that is something that not everyone necessarily wants to do. There's a lot of companies and customers who aren't experts in managing that infrastructure, nor do they want to become experts in managing that infrastructure. And so our commercial offering, which is also available today, allows a user to add funds to Builderlab and then use those funds as they want to use our remote agents, so our agents that work in the cloud, as they want to use the inference and models that we give them, or as they use storage beyond the free storage we provide. And so from a commercial standpoint, what we're monetizing here is the infrastructure that we're providing for those that are interested in paying rather than using— hosting it on their own. And one more thing to add, too, is you can also use it for free. And you can use the service we provide for free. So there's 3 parts to it. The first is you can take the app and run it on your own infrastructure. You can also join Builderlab, create an account for free, and get started using it for free. We'll have limits in terms of the total amount of storage, but they're very generous limits. And so if you want to store anything beyond that, or if you want access to our models, you can add funds and use them directly through Builderlab. And the part that we're trying to solve for here is how do you get everything out of the box. We've had to string together a lot of different tools and infrastructure pieces to make it all work. But for a customer, how do we make it as simple as adding $10 and then all of a sudden being able to have these agents that work in the cloud within your workspace and working with your team?

Alison Kosik

>> Do you think this is the trend? even outside Block, is this the trend that you're seeing where there's this collaborative interface happening?

>> I think it's certainly moving towards that. I think if you look at OpenAI Dots and Grok Bot and Claude Tag, more and more companies, and not just the leading model providers, are coupling the model with additional memory and context and tools. And I think we'll continue to see that trend. And so these companies— all that's to say that these companies are increasingly sort of allowing you to create a perfect agent for you. And I think the next step is then figuring out how do you make that agent work with you and with the rest of your team. And that's where I think the multiplayer piece lands. And I certainly don't think— I would love for Buzz to be a platform that any one of those agents can also participate in. And I think that's where we really help support the ecosystem through this multiplayer collaboration space. But we certainly don't need to be the only place where you can create that agent.

Paul Gillin

>> How are you thinking through the issue of agentic sprawl? Because every company, every software company now has multiple agents proprietary to their platforms, but all being used at the same time. How will this— how does this get sorted out eventually? How do we get control over all these agents?

>> Yeah, I think one piece is on the cost efficiency side of things. The more agents you have and the more you use those agents, that's going to run up your costs. And I think that goes back to why it's really important for us to make sure that Builderlab is this model agnostic tool. Your agent can have a Frontier model or it could also be an open source, much cheaper model. And so the more we offer that sort of selection, I think the more intentional users can be around what they want that agent to have. I think the other piece is we've seen different ways of using agents. So some firms will have, a marketing agent that everyone can talk to. Others will, like Block will have— everyone has their own individual agents. So there might be 50 different marketing agents and that's okay. And the reason why that's okay is Buzz is, the thread or the channel in Buzz is all of the context. And so it doesn't matter if there's 5 agents or if there's 10 agents. If they're all accessing the same conversation and the same context, you can sort of go about how you want to solve the problems with your preferred agent or however you see fit. I think the parts that I would love to work more on is how do you add more guardrails to prevent those agents from endlessly having a conversation? If you ask two agents to have a conversation or have a debate, they'll have a debate and you might be spending more funds than you wanted. And so how do we make that clearer? How do we allow companies to add controls for that kind of thing so that you really get the highest ROI on your AI spend?

Alison Kosik

>> Do you see Buzz eventually replacing some of the collaboration tools that companies rely on today, like Microsoft Teams or Slack?

>> We've really focused on Buzz solving the problems that we have internally and being the collaborative workspace for us. I think a big part of that is, how do we build the right interface for the right role or the right task. And to Jack's point earlier, I think we even need to question the assumptions of what— how do we operate in this new AI world? And so I think there's a lot of rethinking that we're doing both on the tools, but also how we use those tools. And I think really what we're focused on is making Buzz the best tool for us to use at Block and having a plugin ecosystem that allows anybody to make it their own.

Paul Gillin

>> So what's next? It sounds like you're focused on your internal use, but you also want to have a commercially successful product. So how are you going to walk that line of keeping it focused on what your people need to do while also raising awareness in the outside world?

>> Yeah, it's a great question. So what we launched today with Builderlab is a replica of what we use internally at Block. Not just Buzz, but also the ecosystem of tools to manage the authentication piece, the apps platform, and remote agents. And so what I'm really excited about is the products that we're offering external customers is the product that we're using internally. And as we improve that internal product, we're also improving the external product. A lot of the improvements we've been making to the Buzz app, most of the improvements go out to the broader open source community too. And so it's this really beautiful thing where you work on one thing that you use every day, and yet there's a pretty good mechanism for making sure that those improvements go to everybody that's using the tool regardless of whether they're paying or using the open source version.

Alison Kosik

>> Well, Lucas, thanks so much for stopping by theCUBE. I know it's a really busy day for you, launch day.

>> Thank you so much. Super excited to have you here.

Alison Kosik

>> Great. And thank you. You are watching theCUBE, the leader in live technology coverage and in-depth expert analysis. We'll be right back from San Francisco.

Lucas Isaza, Block

(INTRO)

Alison Kosik

>> Thank you for having me.

Alison Kosik

>> It's an exciting day, isn't it?

>> Yes, very exciting.

Alison Kosik

>> It's a long time coming, I'm sure.

Alison Kosik

>> For someone who's never heard of Buzz, talk me through what is it and how is it different from the AI tools that most of us already use and know about?

Paul Gillin

Alison Kosik

>> —can you take me through a typical workday using the tools that we're talking about?

Paul Gillin

>> The architecture is customizable, it's plug-in. The plug-in aspect is a really interesting component of it. What was your thinking in designing the agents to be so configurable and extensible?

Alison Kosik

>> Have you found that you've been able to perfect Buzz just by watching how people interact with AI agents and maybe seeing their frustrations or seeing the shortcomings?

Paul Gillin

>> So you're also open sourcing the technology. I know that's an important cultural dimension at Block, but commercially, what is the thinking behind open sourcing Buzz?

Alison Kosik

>> Do you think this is the trend? even outside Block, is this the trend that you're seeing where there's this collaborative interface happening?

Paul Gillin

Alison Kosik

>> Do you see Buzz eventually replacing some of the collaboration tools that companies rely on today, like Microsoft Teams or Slack?

Paul Gillin

Alison Kosik

>> Well, Lucas, thanks so much for stopping by theCUBE. I know it's a really busy day for you, launch day.

>> Thank you so much. Super excited to have you here.

Alison Kosik

>> Great. And thank you. You are watching theCUBE, the leader in live technology coverage and in-depth expert analysis. We'll be right back from San Francisco.