Introducing Yolo Systems
Introducing Yolo Systems
We’re a motley crew of seasoned technology pros, software engineers, and machine learning experts. We’ve held senior positions at megacap technology companies, and we’ve been founders of early stage startups. Above all, we’re builders. We founded Yolo Systems so we could focus on building tools that would delight developers and engineers building software in the age of AI.
Our journey began in the late summer of 2025. We realized, after several years of building AI-powered development tools ourselves, that AI had become a vital, irreplaceable part of our coding process. Some of us were using it for “vibe coding.” Others were letting multiple agents loose on complex codebases. Still others were running AI coding assistants right in the IDE. In fact, most of us were doing all three! The experience was compelling, and we were increasingly excited about the new crop of development tools and cloud services designed specifically for building software in the AI era. We founded Yolo Systems as our way of contributing to this amazing new ecosystem.
As we integrated AI more deeply into our software development process, we noticed a couple of things. First, despite the fact that a good chunk of our code was being written by AI, we were still working in various cloud consoles and command line interfaces - a LOT!. The AWS Console in particular was one that we were spending significant time in, mostly to get details about resources, and to check on the status of the infrastructure-as-code deployments that run routinely as part of CI/CD processes. We’re big fans of Kubernetes, and the K8S experts among us realized they were running “kubectl” commands hundreds of times a day. So we saw an opportunity. What if we could reimagine cloud infrastructure management - from ad-hoc provisioning of resources to fully automated CI/CD - in the age of AI? If we were to start with the first principles of cloud - eliminating undifferentiated heavy lifting, on-demand provisioning of IT services, and representing infrastructure-as-code (“IaC”) - what type of tools would we build today with the power of LLM’s and generative AI?
Clearly, those tools would not be the AWS Console, or command-line-interfaces like kubectl. For all of us at Yolo Systems, these tools have dominated our professional lives for the better part of two decades. They’re trusty and reliable and they “just work.” But they all have their own user experiences, and passing data from one tool to another often involves additional scripting or annoying and brittle glue code. So we started experimenting with some new user experiences, such as using natural language exclusively to query and mutate resources running in AWS. These efforts commenced in exactly the way one who’s been building software with AI would expect. We defined some tools and MCP servers wrapping the API’s that we call a lot, and hooked those into AI coding assistants like Cursor and Claude Code.
While the initial results were impressive, we wanted a different experience. We wanted the experience of “vibe coding,” where we’d have extended back-and-forth dialog with the agent as we deployed new infrastructure and modified existing cloud resources. No matter how skilled we were at IaC - and most of us have a decade or more of experience with frameworks like Cloudformation and Terraform - we always seemed to have a need to execute some tasks in an ad-hoc manner. This was especially true when bringing up new architectures for the first time. It's just not efficient to iterate in IaC. So we kept building and experimenting and eventually we realized that what we really wanted was an experience similar to the Claude Code CLI that was optimized for managing cloud infrastructure. What we really wanted was to be able to “vibe with our infra.”
So we couldn’t be more excited to launch the Yolo CLI into preview, with support for AWS (GCP will be coming soon). We had told ourselves all along that we knew this tool would be ready when we all started using it instead of the AWS Console. Well, that day has arrived! We’ve been using the Yolo CLI for all our infrastructure work for a few months now, honing the experience along the way until it became a tool we always open in a terminal when we fire up our laptops for the first time on a fresh workday.
We realize that developers have a wide range of AI dev tools to choose from. We think the Yolo CLI will appeal most to those that really like the experience of the Claude Code CLI. We’ve designed this user-experience to be CLI-native, and we’ve added in all kinds of features that make it a dream to use. The Yolo CLI has a rich inline command menu and shortcuts for selecting specific data items in the context. While you can see the JSON output of the API calls it makes on your behalf, the data is displayed in beautiful formatted ASCII tables. It supports memory so you can pick up where you left off in a given session. It has a robust permissioning system so that the agent won’t do anything you don't allow it to do. You can easily add local and remote MCP servers to enhance functionality. Best of all, your cloud credentials never leave your local machine. All authentication is done through the existing mechanisms you already use to connect to clouds like AWS.
We find that we really love using the Yolo CLI for our own work and think that you will too. We’re a tiny team that is entirely self-funded and we plan to keep it that way. But we value your feedback and will do our best to incorporate it as we continue to iterate at a fast clip. Be sure to join our community on Discord and keep an eye on this blog too, as we’ll regularly be posting videos of the Yolo CLI in action as well as updates about new features and capabilities. We’d love for you to accompany us on our journey as we reimagine cloud infrastructure management.
After all, you only live once!