Ship 26 London - Keynote

[music] Hi everyone. Thank you. Thank you. Hi everybody. It's so great to be here. Since ship last year, the world has changed. We're writing less code by hand, but more ideas are coming to life than ever before. Six months ago, less than 3% of Versel's deployments were triggered by coding agents. Since then, that number has grown 17 times. And over half of deployments on Versell now come from agents. But the bigger shift is what those agents are deploying. Since the beginning of the year, agentic workloads on Versell have doubled. volume on AI gateway grew from two trillion to 20 trillion tokens per month. And the shift we're seeing is that we're using agents to deploy software that can think. We're shipping agentic sites and apps. We're using agents to ship agents. This is a profound change. So I want to go back in time and show you how we got here. Because our new world is agentic. But it all started with websites. I remember building simple sites with HTML, CSS, and JavaScript as a kid, as one does. I was so excited to put my creations online. And that's what the web is all about, sharing our ideas with the world. We built the first version of Versel to make that easy for anyone. It was infrastructure for pages, caching, and global content delivery. Today, we serve the fastest, most beautiful websites in the world for companies like Co-op, Paul Smith, and Hely Hansen. But the world is dynamic and as the web grew, sites turned into data applications. So, Verscell built infrastructure for servers, APIs, and databases. Now sum up Trip Advisor and your very own London Northeastern Railway run apps at scale on Verscell. For over a decade, we've been building Verscell into a cloud where you can run everything. Our vision is to build a true full stack platform. And in the last year, we've shipped some huge steps towards that vision. We brought on some of the best finest Python developers in the world to extend our core infrastructure for back-end frameworks. You can now run backends like fast API, Flask, Express and Hono at scale right on Verscell. We built out a complete agentic layer. You can now run longer duration functions on fluid compute, make workflows durable, and spin up secure sandboxes. You can host your MCP servers and make your app available to any agent. And we added the best databases in the world to our platform. You can manage Amazon Aurora, Aurora DSQL and Dynamob along with hundreds of highquality integrations like open search all through Verscell. This means you can run any architecture you want on our infrastructure. Right here in London, Octopus Energy runs a Nex.js front end with a Python back end, helping millions in the UK monitor their power usage. So in addition to frontends, you can host backend only services on Verscell like REST APIs written in TypeScript and Python. You can deploy workflows that handle long running asynchronous work even with a human in the loop. And you can host cues that process billions of messages per month. But developers tell us it's still too hard to wire all of those pieces together. So, we fixed it. Today, I'm excited to announce Versel Services. It's a developer experience you know and love for your full stack application. You can now develop front end and backend together with one command vcde dev. Everything spins up locally and when you push a commit, you get a preview URL for your entire application. Not just the front end. Even backend only commits generate a full preview you can test before you ship. And all of the services you deploy can talk to each other privately without ever touching the yucky public internet. You can now run all of your microservices, I said microservices, microservices on Verscell and everything just works. But in the agent era, these services are becoming autonomous. Websites and apps used to respond to user input with logic. But now they have agents inside that can understand intent and take autonomous action. Every new generation of software demands a new generation of infrastructure and the agent era is no different. Agentic software needs agentic infrastructure and that's exactly what we're building out of Verscell. Agentic infrastructure has three parts. One, Verscell is where coding agents deploy software. When you ask cloud code or codeex where to deploy, you get Verscell because Verscell is built for the way agents work. Second, Verscell is where you build and deploy your own agents. We give you every tool you need to build and run apps and agents in production securely at scale on one platform. And third, Versell itself is automated by agents. Verscell runs your apps in production, of course, handling traffic, traces, observability, and anomalies. And so that data gives our agents the context they need to investigate autonomously and then surface poor requests, not just alerts. For context, teams using clot code deploy to overcell five times more frequently than teams who don't. That's because we set the standard for developer experience. And now we're doing the same for the agents that those developers use. Coding agents love deployment overcell because we give them everything they need. In fact, when you ask your agent to verify its work, it needs a live URL it can test, right? So, Verscell gives every deployment a dedicated secure preview environment. When you ask your agents to ship experiments, it needs to roll them out safely. So, Versell gives every experiment a feature flag and gives you control and confidence with instant rollbacks. And you don't want your agent wasting time clicking around a dashboard. An agent is most efficient when every part of the platform is available in its own language. And that's what Versail gives it through our CLI, API, and MCP. This year, an engineer at Meta opened Claude code as one does and asked to build an internal tool and it was when it was time to test it, he asked Claude where to deploy. Lo and behold, push it to Verscell. A week later, everyone on his team was deploying Versell and within a month Verscell was the go-to platform for Meta Super Intelligence Labs. Meta.ai, AI Meta's Frontier AI product was born on Verscell and that happened even though Meta had already built their own deployment platform for decades. But even the most powerful infrastructure doesn't natively speak the language of agents. Versel does agentic infrastructure is a platform we built, but what matters the most are the products you build on top of it. So, I want to invite Tom Aino to the stage to show you how. Thank you. Thank you, G. Thank you so much, everyone. So great to see you. It is so great to be back in London. G just told you why coding agents love Verscell's infrastructure, why they choose to deploy to Verscell's agentic infrastructure. But what about when you want to build your own agents? Versel's agent stack gives you all the tools you need to create and ship your own agents. Agents need context. Uh agents need to connect to models, execute complex workflows, and connect to data and apps. Let me walk you through these tools now. Starting with how to connect to models. First up, we have the AI SDK. Have any of you used the AI SDK? Show of hands. Lovely. AI SDK was initially released three years ago this week and it's been in active development ever since. It's the universal toolkit for building AI frameworks, apps, and agents. It's platform framework and model agnostic, and it allows you to generate text, images, speech, video, and more. It also makes it trivial to add things like streaming and fallbacks to your projects. The AISDK is now downloaded over 15 million times per week and has become the standard way to access any model from both from any provider in both Typescript and Python. Shout out to the AI SDK for Python. It's being used at massive scale by companies all over the world like B. Brex helps businesses like Verscell with corporate cards and spend management. They run expense audit agents that review thousands of transactions at a time. Those agents call models, invoke tools mid-generation, stream results as they come back, and parse structured financial data to surface anomalies. The AISDK simplifies all of this and provides a standard abstraction layer. One of the things we love is that you can experiment with different models for different parts of your application and try out new models as they're released, all without needing to make any changes to your product code. You get to focus on building your features and the AI SDK handles the rest. Okay, so AI SDK makes integration really easy, but you still have to actually manage the connection to the underlying providers and to the models that you want to use. And that's where the next tool in our agent stack comes in. The AI gateway. AI gateway provides a unified interface for accessing AI models. But it's actually so much more than that. Internally, we've been calling the AI gateway a token delivery network. And I want to explain what I mean by that. In the early days of the web, we had something called the hot origin problem. Popular sites would suddenly receive spikes in traffic from everywhere. But content lived in only one or a few origin locations. This created overloaded servers, slow downloads, and an unpredictable user experience. The web sort of outgrew the idea that every user should fetch assets directly from the origin. And so the CDN was born. The CDN became the internet's performance and reliability fabric. Distributed edges, intelligent routing, failover, observability, centralized policies, and more. AI use cases have now outgrown the idea that every token should be fetched directly from the model provider. Tokens have become a production dependency and model labs are the new origins. They're powerful for sure, but they're also expensive, rate limited, and geographically and operationally variable. And as our friends at the model labs know, this is a really hard problem at scale. So that's why we built the AI gateway. It serves tokens through the same global network that Verscell has run for over a decade. It routes around failures, simplifies off and centralizes usage with things like spend tracking and granular observability. And of course, it offers zero data retention, which is particularly important to our enterprise customers. AI gateway lets teams operate AI traffic as infrastructure instead of bespoke provider integrations. Now, serving tokens reliably is one of the important problems that the gateway solves, but the other is about model choice. Today's agents don't actually use a single model architecture. They use many models from many providers. In fact, our AI gateway production index showed that teams running agents at scale route across 35 different models. That means routing is no longer a feature. It's actually how you run AI in production. One of our customers, AKQA, built a chat GPT app for Starbucks using uh Nex.js and the AI gateway. It's an interactive bot that helps people discover new drinks they might like. They use different models for different tasks to balance power and cost. Gemini Flash handles retrieval and summarization when someone asks a question. Then recommendations are handled by GPT5. And even more models are used to generate recipes all with a single AI gateway API key. AI gateway is serving over one trillion tokens per day. It gives developers simplified access to hundreds of models from dozens of providers and provides powerful observability. It's actually a huge part of what makes Verscell the open platform for AI. Okay, so Versel serves pixels and tokens instantly and reliably. But modern software doesn't follow a simple request response model anymore. Agents may need to run for hours, even days, and across many complex tasks. That's where the next tool in the agent stack comes from comes in the workflow SDK. Longrunning workflows and backend jobs are nothing new in software, but they're one of the most challenging and annoying things to build. Many things can go wrong and failures do happen. Timeouts, dropped connections, lost state. Without a primitive for durability, you're handstitching things like retries and state persistence. It can get really ugly. That's why we built the workflow SDK. We like to say it provides infinite compute durability. It enables you to build longunning apps and agents that automatically suspend, resume, retry, and maintain state with ease. Door Dash uses the workflow SDK to run traditional ETL jobs, making sure that all the data in their app is always up to date. And our customer Flora, sorry for the spoiler, a minute ago, uh built an entire AI design platform on top of the workflow SDK. Their platform helps designers generate visual content at scale. Inside it, agents fan out across 50 different image models to generate visual directions from a single creative brief. Workflow SDK checkpoints every step of every agent and pauses when the job needs human input. And because every failure is automatically retrieded, the designer never needs to start over. Now, the whole idea behind agents like this is that they can solve complex tasks across multi-step workflows. And one of the primary ways that agents try to accomplish most tasks and what they love the the way that they love to accomplish most tasks is by writing and executing code. And that's where our next agent stack primitive comes in. Versell sandbox. While LLMs have gotten really good at producing working code, that code is still untrusted. And we can't run untrusted code in the same environment that has access to our production systems. We need a special production grade environment that's designed for executing code in a secure and isolated way. Verscell is actually no stranger to this problem of untrusted code execution because of preview deployments and builds. We host over 1 billion production grade preview deployments and every single day we run over 6 million builds. Every one of those builds happens inside an isolated microVM compute environment. With Verscell sandbox, the same isolated compute primitive is now available to you and your agents. Verscell sandbox is built on top of fluid compute. It supports dynamic real-time workloads for agents, code generation, and developer experimentation. Each sandbox is a fully functional computer with a file system, security boundary, and even full Docker support. When you spin up a sandbox, you can install packages, run containers, and even configure Reddus or Postgress as test dependencies. It's already being used in production at scale by the best companies in the world. One company in particular that we use every day at Verscell is Notion. Millions of teams use Notion as their AI workspace, capturing knowledge, answering questions, and moving projects forward. And developers can extend Notion agents with custom code, syncing CRM data, turning Slack threads into content, or connecting to the workflows their teams already rely on. But that code has to run safely. And that's why Notion custom agents run on Verscell sandboxes. Since each agent gets its own generalpurpose compute environment, developers can build almost anything they can imagine and everything outside the sandbox stays protected. So now our agents can execute code securely inside longunning workflows. But in order to do anything useful, our agents need access to data and tools, the apps that we all use every day. And that's where the next layer of the agent stack comes in. Verscell connect. Verscell Connect is a brand new building block that allows your agent to securely connect to all the data and systems they need, including the apps and tools they use to communicate with users via secure, short-lived, and minimally scoped access tokens. I'm excited to invite updo mentioned agents only truly become useful when they have access to the systems that and tools that your uh business uses. Let's say we wanted to build an agents that automatically processes call transcript for our sales team. We might want an agent to also update statuses in Salesforce to reflect what was discussed in the sales call. It could also extract feature requests from the call and then create issues in in systems like Linear. And of course, our sales team wants to interact with their agents from Slack. This sounds like a basic agents for processing text, but it's actually a complex workflow with real security implication. When you build custom connections to for your agents into Slack, Salesforce, and Linear, you have to provision and manage access tokens. And most systems provision longlasting credentials, meaning the agent effectively has permanent access. What's worse, those credentials are usually scoped to a human that created this access token, not the agents. So this means your agents have the ability to perform any action that you can, not the actions that they're designed for. That's why we built Versel Connect to solve all of these problems. Versell connects give your gives your apps and agents secure, short-lived, minimally scoped access to your system. It's also integrated into Briscell's observability suite so that you can see how those tokens are being used. So, let me show you how it works by setting up the first connectors for the agents I just described. Here inside the Versol dashboard, I'll start by creating the connectors that I want. I'll click on create connectors. And as you can see, we have a growing list of built-in connectors already available. It's including Slack, GitHub, Snowflake, Linear, Salesforce, and many more. And even if you don't see a connectors that we don't yet support, you can create your own using oath and API keys. Let's start with Salesforce. I'll use the workspace that I want my agents to have access and then hit create Salesforce connector. So what just happened looked simple but behind the scene vers is taking care of a lot of complexity on the integrations. So normally connecting to Salesforce means you have to deal with oath work oath flow token management credential storage and refresh logic. But with connect all of these are taken care for you. This connection issues shortlived scope tokens on demand when the agents really need to perform action. So let's start creating the linear connector here. I'll go to uh to uh create connector. I pick linear. I make sure I have the right workspace. And what happened is that versel connect enables me to choose exactly what permission I need to give to my agents. In this case, the agents need comment to comments and also create issues for uh the agents I described. So I'll only give it those permission and then simply create the connector. So by restricting our agents to those permission and scope up front, we can make sure that this agent only performs the task that we want it to be and still have the powerful automation. So I'll finish creating the connector and I'm landed back in a details page. So you can see the project creation here. You can install uh you can see the installation trigger and you can also see the usage data later. Uh let's quickly test our token here. So, I'm going to test the user token. I'll click on test user token and I'll authorize my APIs and then I'll authorize on the linear side and great, I'm given back with a token that my agents needs to run the specific task. It has the expiry. It has the name and let's start. Let's create the last connector that we want for our agents. I'll select Slack and I'll make sure that I have authorized uh workspace and I hit create. I'll um install it later for uh a little bit. I'll I'll uh I'll I'll show you that. And then going back to the page. So now that we have the three connectors installed, our agents now has access to everything it needs and it never stores long live credential. Back to you Tommo. Thank you. [music] With the with Versel Connect, your agents can securely connect to your full range of internal systems, including your CRM, ERP, HRIS, data warehouses, collaboration tools, and more. But I want to double click on collaboration tools. This is how you interact with your agents and how your agents want to be able to interact with you. And that's where the next primitive in the agent stack comes in, which is chat SDK. The most powerful agents we've built at Verscell are effectively co-workers and we interact with them where work happens as it's happening. For many of you that might be Microsoft Teams or Google Chat, but for us overwhelmingly that's Slack. We believe that powerful collaboration tools like these are the primary way most humans will interact with agents. But there's a problem. Even though these tools share a common set of core capabilities, they all have vastly different APIs and interfaces. Chat SDK provides an elegant abstraction layer enabling you to target dozens of apps. With just a single line of code, it enables your users to interact with agents across tools like Microsoft Teams, Google Chat, Slack, Discord, GitHub, Linear, Telegram, WhatsApp, and so many more. Nano Claw helps companies run AI agents and they built their platform on the chat SDK. It's a single codebase, but it delivers agents across 15 different messaging apps. So that's the Verscell agent stack. It's a singular set of endto-end capabilities covering everything needed to ship production agents. We've taken everything we've learned building agents over the past few years and turned those learnings into best-in-class primitives that work at Verscell scale. We love them and our customers do too. The stack is powerful and fills a real gap in the ecosystem and we believe every one of these primitives is bestin-class. But there's still a lot of complexity. Each of these primitives still needs to be wired up into a single cohesive agent. But what if they didn't? As we've built out the agent stack, this is the question we kept coming back to at Forcell. We don't just want to build the world's power most powerful agents and we don't want to just enable you to build the world's most powerful agents. We want building those agents to actually be delightful. And that means not just providing the world's most powerful agent stack primitives, but also the world's easiest and fastest way to stitch those primitives together. To make that possible, we built the newest member of the agent stack family. Welcome, Eve. Verscel's framework for building complete end-to-end production agents. Now, I have so much to say about Eve, but before I get into that, I want you to see it in action. So, I'm excited to invite up Char, product lead for Eve, to show you how it works. Thanks, Tommo. Over the last year, we've taken everything we've learned about building agents and we packaged it into Eve. Let me show you how it works. I'll get started with a single command in my terminal. This command will scaffold an agent directory, install dependencies, and start an interactive chat session with the agent. Next, I'll configure our model provider. I'll use Versel AI gateway. Select a Versel team and link to a project in that team. And we're done. Now, let's test it. Who are you? Okay, that was fast. I just built a fully functional agent in less than a minute. [applause] Now this agent runs with just two files. Agents.ts agent.ts which defines the model and instructions.mmd that defines the agents identity. And that simplicity is what makes it so easy to build agents with Eve. Now let's build a real use case. Hi just described a sales agent that processes call transcripts and updates Salesforce and linear. Let me show you how simple it is to build that agent with Eve. First, I'll give the agent an identity. In the instruction markdown file, I'll describe a go to market assistant that can create linear issues and update Salesforce opportunities based on call transcripts. And just like that, the agent has a new charter. Next, I'll give it tools it can use to take actions. The agent need a linear tool so it can create issues and tools go inside the tools folder and the file name is the name of the tool our agent sees. Let's define a linear tool. The description is what the agent reads to know when to use the tool and the inputs are the data points that the agent has to fill in like the title and the description of the issue. We can uh use the linear connector that Hedi set up earlier and then implement the linear API for creating an issue. I also want the agent to know how to create a linear issue. So I'll add a skill and skills go to skills folder. This skill tells the agent that it should look for feature requests in the transcript and separate linear issues by topic. Great. Now let's test the agent. It should be able to create linear issues. It's going to load the skill. Create the issue. Great. The issue is created and it should also show up in linear. Next, the agent needs to update Salesforce opportunities. We want to use Salesforce CLI to make updates, but that means our agent needs a shell and we want the commands to be executed safely in isolation. So, I'm going to configure a sandbox that the agent can use to make updates. This sets up a private VM for the agent to work in walled off from everything else. I'll create a sandbox.ts. This is going to configure the sandbox and then install Salesforce CLI in sandbox bootstrap and also lock the doors. So the sandbox is only allowed to talk to Salesforce and nothing else. I'll give the agent the Salesforce tool just like we did with linear. But Salesforce updates need approval from someone on the team. So I'll add human in the loop. And this tool will always require the agent to ask for human approval. We'll use heady salesforce connector and last the tool execution function that will call Salesforce CLI and make updates to the opportunity record. Now let's test the Salesforce step. I'll tell it to update Salesforce opportunity to closed one. That's human in the loop and done. If you go back to Salesforce, refresh this page, it should say closed one. Awesome. Now I need to make the agent available in Slack for the sales team. So I'll run slash channels. Select Slack. Yes, I want the Slackbot. This is going to install the Slackbot in the workspace. Deploy to Versel. And done. Let's give it a test. I'm going to tag our agent. And that's it. We have a live production ready agent that our team can use in our collaboration layer, Slack. But we're not done yet. The agent is also fully observable. Let's take a look at the conversation we just had with it in Slack. So in the Versal dashboard, click in observability, agent runs, and I can see the entire conversation history, token usage, length of the run, inputs, outputs, and the agents reasoning. I can even drill down into each of the tool calls that the agent made. And that's Eve. We just filled a production ready agent in five minutes. Back to you, Tommo. Thank you, Char. Eve truly feels like the missing Next.js for agents. It's opinionated based on everything we've learned over the past several years, but more importantly, it's opensource and modular. It's built to work seamlessly with Verscell with high cohesion to our infrastructure, but it's completely customizable, so you can make it your own. You can swap out providers and keep the durable foundation underneath. Before Eve, every agent needed its own scaffolding and integrations. With Eve, every agent is just a directory laid out the way you already think about your code. Building an enterprisegrade agent used to take weeks. Now, as you just saw, it takes minutes. Eve is going to change the way many of us build agents. It's already completely changed the way that we build agents inside Verscell. But next up, I want to invite Malta to tell you about one of his favorite agents we've built um that many of us are very excited about. Thank you. Thanks, Domo. Uh, I'm Malta Versell CEO. The agent Tommo is most excited about, it's mine. We built it for Versell first and now you can use it too. It's called Versel agent. Whoa, wait, wait. Hold on. Shoot. I'm actually getting paged. Um, okay. Wow. Okay. Vzero is having a partial outage. Um, let me actually get to my laptop. Imagine I have a laptop and check this out. All right. So Versell agent already investigated this. Let's let's dive deeper. Okay. Okay. Okay. It found that the API key object is undefined at runtime. The errors appeared after the last deployment a few minutes ago. It recommends an instant roll back to the previous release. That's actually really smart because when in doubt, you always want an instant roll back as the right action. Let me approve this. Now with my permission, Versell agent is rolling back the last production deployment before the 500 errors showed up. All right, the roll back was successful. Vzero is back online. Versell agent will now start working on a fix. Um, but we don't have to wait for that. All right, let's start over. Hey, my name is Malta. I'm REL CTO. Versell agent is great because it autonomously detects incidents, only escalates when there are true positives, and has the analysis already done by the time I get to my computer. Now, I do realize that the first question every [clears throat] CTO in this room will ask is, how is it safe to let an agent do this? It's the right question because most agents inherit the users permissions. They run as you and they can do everything that you can do. A single bat prompt has your full blast radios whether it comes from you, a teammate or confused sub agent. Versel agent has a firstofits-kind permission model that combines plan mode with granting permissions. Rather than asking you to approve actions one at a time, Versell agent plans what permissions it will require to complete a task and then ask you to approve them in a single coherent step to roll back a broken deployment. Versell agent only asks for temporary permission to perform instant rollbacks like we just saw. If it needs to purchase stale cache, it asks for temporary cache access to that project and nothing else. Every agent, every action runs in an isolated sandbox before it touches production and anything that changes production state waits for a human to approve it. The agent never has more access than it needs for the task it's doing, but it also doesn't constantly prompt you for more permissions. A great balance of progress and safety. It's the part I'm most proud of and also why I don't worry that Versell agent will be the cause on my next 2 a.m. page. Wait, wait. What's What's that going on? Someone not turn off their phone. It's 205. Oh my god. It was me. It was me. All right. Oh, I got a text from GMO. Dude, you perfectly know I'm on stage right now. Anyway, I realize you all realize this is staged, but let me tell you, gexting me to complain about something not going right on our website is about as lifelike as things get. So anyway, let's copy paste that message into Versell agent. Actually, let me add make no mistakes just for good measure. Um, all right. Versell agent is now looking at speed insights for our website to see if anything changed. Oh, wow. Okay, there's a 4 second LCP. Something is majorly wrong. That's not That's not good. Now our cell agent is going through a uh going to review all the recent changes to narrow down the problem and it found that we added an await on the fabicon route. It checks performance on that endpoint and it confirms that the an update caused the problem. It finish the investigation with a root cause analysis and it looks like I'm going to have to talk with Matan. He's here in the audience somewhere. No worries. The good news is the Versell agent can write a fix and open AP APR. When I approve it, the change will go into production. Okay, back to G fix the chief. Uh zero lines of code written today by me. All right, those are two examples of how I can of I use how I use forcell agent but can help you manage anything on the platform. When you ask to fix your build, it reads your deployment logs, finds the failing config, validates the fix in the sandbox, and redeploys with your approval. Ask why your build got slower in the latest deployment and diffs the build against the previous one and tells you what changed. You can ask it to find the top accessibility issues in your project and will run a review and open the PR. And if you ask it to fix the 500 showing up in your logs, it will do the same. But what I love most about with about the agent is that you don't have to ask. Versell agent sees your app running in production. So when traffic spikes or an alert fires, it will investigate immediately and bring answers to you just like it did for me when I got paged here on stage. The examples I gave you were normal infrastructure and performance problems, but soon agent will be able to call specialists. It will it will run deepseec for deep security review across your whole codebase or inspect your front end for design and UX quality. It's one agent that understands your stack and your infrastructure with exports it can call on demand. Rell agent is available in private beta today. Scan this QR code to request access. Uh oh, not again. Great. It's our CEO Jean. All right. Uh okay. She's writing, "I'm panicking a little bit. Did we actually flip the feature flags for everything we announced today?" Now, obviously, I could now go like lock in the dashboard, look at versel flex myself, but what year is this? 2024. Let's ask for cell agent. What did we ship today? All right, we have enables services. G talked about this. Enable Eve. I'm so excited. I can't wait for all of you to try it. Enable enable Versel agent beta. You know, just talked about it. and then enable Versell password. Actually, what is Versell passport? Uh, we haven't any heard anyone talk about that yet. I might have just leaked the rest of the keynote. Anyway, to talk about Versell passport and other things, please welcome on stage Jean. Thanks, Malta. Verscell agent shows you what's possible when you build agents the right way. And every company in this room is going to build an agent just like it. There are two types of people in this room hearing that. Some of you are sitting here thinking let's go. I am shipping that tonight. Then there are the CIOS and CTO's in the room who are thinking ah no because you can already feel what's coming. Shadow agents writing to systems with no audit trail. AI bot user closing tickets spend you can't explain. Both of you are right. Building agents is easier than you think and way harder than you think. I'm going to tell you about that tension and what we've learned at Verscell by living it. Drew Brevik, who works for me, is head of go to market engineering. In June 2025, he had the sexiest job in the world. His mandate, build the agents that transform how Versell goes to market. It worked. A year later, agents are part of our daily workflow running at scale across our entire goto market organization. You did a great job, Drew. So today, I'm handing you a pager. Let me tell you why. What we learned was that agents are free. Free as in free puppies. Everybody loves puppies, but they pee on your floor. They eat your furniture and you can't go on vacation. Agents are free because anyone can prompt Claude. But agents are software and we all know that software is never done. Someone has to maintain them, update models, and build new features. Building hundreds of agents taught us hard lessons. First, we saw the same problem solved over and over. Multiple agents needed to connect to the same internal systems. Each team built their own integrations from scratch. Second, each agent was reading from different knowledge bases. Our team would ask the same question and get different answers. Third, we had no visibility. No one knew how many agents existed, who built them, or what data they touched. Fourth, that lack of visibility also meant adoption chaos. I'm on Slack, I type at 500 agents pop up. I don't know what any of them do. And last, we also learned that chat isn't all you need. We started with the idea that Slack was the universal interface, and that was wrong. The agents that actually got used also had frontends for permissions, for visualizing data, for workflows, and for keeping humans in the loop. We learned that all of our agents had to work on day one and on day 100. And I'm happy to tell you that they do. We run over 100 agents in production at Verscell and they're part of how we operate every single day. I want to tell you about 10 of the most important ones and the order matters. We started with the most obvious use cases and worked our way towards the agents that changed internal processes and transformed how we operate the go-to market team. Vertex is our customer support agent. It resolves over 91% of Versel's support tickets across the help center, Slack, and Docs chat. Deal one is our deal intelligence agent. It listens to every sales call, coaches our reps in Slack, and runs a post-mortem on every lost deal. The Deal One MCP has been called 17,000 times this month. Draft Zero is our content agent. It writes the first draft of every blog, change log, and customer story we publish. Az is our AEO agent. It tracks how Verscell's brand and content shows up across AI search every day. It runs hundreds of prompts across dozens of coding models. Ravoa is our Salesforce update agent. It pushes critical record changes into Salesforce with a human in the loop. It saves 9 hours of time for our RevOps team every day. Penny is our finance and ops agent. It has access to our billing platform, payment provider, and monitoring systems. It saves our finance and on call engineering team hours answering billing tickets. And next, there's V. V is our routing agent. It routes requests to all of the other internal agents. Remember that adoption problem? We still have 100 agents, but V is the front door to all of them. Ask via question and it picks the right agent for the job. The last three I'm going to show you are the ones I want to spend real time on because I think every single one of you is going to want to build one of them. First, DZ, our data analyst. DZ gives our entire company 247 on demand at analytics and data science. Anyone at Verscell, engineers, AEES, finance, support can run analytics on our data from our warehouse without filing a ticket or waiting on the data team. Users can ask simple questions like how many leads we got from a campaign. And DZ writes and runs basic SQL. But DZ is also a data scientist. If users need statistical analysis done, it spins up a sandbox and runs Python to generate reports. DZero is the most used internal tool at Verscell. It answers 30,000 questions a month and it's safe at scale. DZ doesn't run in god mode. Every query is scoped to a user the user's permissions. If you can't see a table in Snowflake, DZero can't show it to you either. Under the hood, DZero needed a semantic layer. And even though you can ask questions in Slack, it needed a UI so people could explore charts and drill into data. Next is Athena, our sales cockpit. Salesforce announced headless. We've been running on it for months. Athena picks accounts, plans outreach, tracks signals, and runs the weekly motion for every AE at Versell. Shortly after it went live, pipeline nearly doubled. Every AE uses it every single day. Under the hood, Athena needed the same semantic layer that DZero needed, plus durable workflows and secure connections, plus a UI because GTM agents are more effective with pixels and buttons than just a Slack channel. And last, Lead Agent, our autonomous SDR. We trained lead agent on the playbook of our best SDR and now it runs the playbook itself 247. You may have seen the headline last year. We took 10 SDRs down to one. That was lead agent. We redeployed those nine reps into bigger roles and raised our quota. We've seen a 32x ROI and it costs $5,000 a year to run. It performs at the 99 90th percentile of our reps and one engineer maintains it part-time. Under the hood, lead agent runs the stack Tommo just showed you. AI SDK, workflow SDK, and chat SDK for Slack. And it's open source, so you can go build your own today. Building these agents was easy because we use Nex.js JS and our frame agent framework Eve and running them was never a problem because they run on Verscell. Remember when I said building agents was harder than you think? The hard part is everything around them. Who can access them, how they authenticate, what data they can touch, and providing all of it to your security team. So we spent the last year building the platform that makes that easy. And today we're making it available to you. I'm excited to announce Verscell for enterprise apps and agents. It's the Verscell developer experience you love for everyone at your company with identity and access built-in and the option to run it in your own AWS tenant. Like all of our products, it's portable because it's framework and model agnostic. And we built this for ourselves first. And now it's a platform you can build on too. Let me show you the three most important pieces. Before today, you had to manually provision and offboard every Forcell user. Enterprise managed users gives you central control over employee access to Verscell and Vzero with a full audit trail of every action they take. I said earlier that anyone can prompt Claude. The reality is your employees are already doing this whether you know it or not. That's called shadow IT and AI has already caused major data breaches in the enterprise. So even if you can control who builds with AI, you still have to limit the access to apps and agents they build. That's why we built Versel Passport. It puts every internal app and agent behind your IDP by default. Internal tools stay internal. Employees can only see the apps they need and nothing is publicly exposed. And you can deploy all of this to your own AWS tenant. You heard that right. You can even run Verscell functions in your own AWS account. Whatever you build with agents stays inside your security boundary. The teams who ship fast, securely at scale are the ones who will win. We built enterprise apps and agents so that can be you. G, back to you to wrap up. Thanks, Jean. Killed it. Today we showed you that Verscell is a platform where you can build and run everything, any kind of website, app or Your back-end frameworks can run at scale. With Versell services, you can develop, preview, and deploy backends and frontends together. With Eve, anyone on your team can build an agent in minutes or seconds. Versel Connect gives those agents secure access to all of your services. Versel Passport makes your internal agents and apps stay internal behind your IDB. And you can run all of this in your own AWS account. You can start today. Ask your coding agent to install the Verscell plugin. Then you can build anything, a website, an app, an agent and ship it at a global scale. and their cell agent will keep an eye on This is agentic infrastructure. We can't wait to see what you ship next.