Redefining Developer Platforms in the Age of GenAI

Over the past 12 years, I have personally witnessed the dramatic shift in developer tooling. I think back to my first job as a developer after I graduated college - my team was one of 15 spread across several companies building components for a distributed system. To say things were "complex" is the understatement of the year.

Simply building the lab environment (3x Dell R920 servers + networking config) and setting up the OS / software took the better part of a week. At one point, my boss had us try out Ansible, SaltStack, and Vagrant to speed up the provisioning process. We reduced that timeline down to about 3-4 hours. That sparked my love for DevOps. From there, we graduated to containerizing our applications (Windows Python apps + Wine to run in Docker), orchestrating our agents via Docker Swarm, and finally automating deployments to dev, test, and prod with GoCD.

Looking back, those things seem trivial. However, modern tools are built on the lessons-learned from those before us. We had to struggle with long builds. We had to "find things to do" while our code compiled (I often waited 4+ hours for my local Firefox build to compile). We had to figure out how to connect all of those disparate systems to make our lives as developers easier. Today we get all of that "for free." Every platform has 50 integrations. You simply need to click a few buttons, and things generally work.

Changing the Paradigm

When I first joined CodeCargo, I was intrigued by GenAI. It was an interesting tool, but it didn't really perform better than your average developer. Honestly it tended to perform much worse. At one point, I asked GenAI to design a highly-available OCP platform architecture on Azure and it started talking about how to load balance traffic between AKS clusters.

Now, things are different. Not only have the underlying models gotten better, but we have improved our strategy and techniques. In addition, people have become comfortable with chat-based interfaces.

One thing I've noticed is that the software industry struggles adding GenAI in a novel way - especially for existing products. It's incredibly easy to add AI to your app and it's just a chatbot that helps users fill out forms. That no longer provides unique value. Companies need to figure out how to build their apps around AI so it can be integrated into core features. This is the only way to provide new value to users.

As we build CodeCargo, we deeply integrate GenAI into key features when it provides tangible value. This allows us to re-examine how users should interact with our platform in a novel way. For example, service catalogs are great because they allow developers to keep track of their applications, configuration, and infrastructure. However, every time I worked with a company that used a service catalog, it was stale because it requires manual updates from developers. CodeCargo will make that process automatic.

By building GenAI into the core of our platform from day one, we are innovating on behalf of our users. We can provide unique experiences and actually empower our users. Every DevOps platform company need to follow this path if they want to provide continued value to their customers.

Redefining Developer Platforms

We released our private beta waitlist earlier this week with a few key features:

• Self-Service Workflows --> every DevOps customer I've worked with wants this

• GenAI Workflow Editor --> we leverage your entire GitHub context to write more effective and compliant workflows

• GitHub Workflow Visualizer --> see what your GitHub workflow looks like before you run it

However, I'm incredibly excited about the future. At the end of the summer, we are going to launch our agentic service catalog that will shift how developers work day-to-day. At its core, our service catalog will combine the power of Generative AI to understand your application's context (code, infrastructure, deployments, documentation) with declarative code to keep your catalog up to date. Goodbye outdated point-in-time snapshots.

For a long time, developer platforms have made a few key assumptions:

• developers will work through manual processes

• developers will manually update systems

• form-based workflows are the best way to automate tasks

• developers will keep documentation up-to-date

These assumptions have served us well for the past decade. However, as a developer, I don't want this to be where we end up. We're building a platform that eliminates a ton of manual overhead that has burdened developers for years. If you'd like to embark on this journey with us, join our private beta.