The AI Code Generation Problem Nobody’s Talking About
When Nitric CEO Steve Demchuk met his co-founders at a FinTech startup, they were building high-end frontends for banks while wrestling with the constraints of regulated environments.
Fast forward to today, and that frustration has evolved into the Nitric framework that’s flipping the script on how developers think about cloud infrastructure.
“The biggest frustration for talented developers is when they have to work against their own architecture,” Demchuk said in this On the Road episode of The New Stack Makers, recorded at Google Cloud Next. “They’re being told to use outdated DevOps tooling. They can’t use the language they want to use. There’s just a lot of handcuffs — that we can help them get rid of.”
In this episode of The New Stack Makers, TNS founder and publisher Alex Williams asked Demchuk about solving the challenges faced by organizations that want to modernize their infrastructure but face many constraints, and how Nitric’s solution lets them automate infrastructure based on app code while using the tools they want to use.
Not Infrastructure as Code, Infrastructure From Code
Traditional Infrastructure as Code (IaC) has developers writing application logic in one codebase and infrastructure specifications in another — a disconnect that creates friction and slows development. Nitric’s approach is fundamentally different. Instead of forcing developers to become infrastructure experts, the platform examines application code and automatically generates the necessary infrastructure specifications.
The inspiration, Demchuk said, came from the users themselves. “They kept telling us, ‘Hey, I love how much faster I’m writing app code, but what I don’t love is I’m still writing all the TerraForm or I’m writing all the AWS, CDK — can we automate that?’”
In response, Nitric’s founders built a tool that let developers simply write application code and then leave it to the platform to infer the requirements and automate the infrastructure creation. The result is what he calls “infrastructure from code” — a paradigm shift that eliminates the traditional separation between application development and infrastructure management.
The Nitric framework abstracts common cloud application elements like databases, compute, storage and messaging across multiple languages and cloud providers. Through an SDK and CLI, developers can write in their preferred language while Nitric handles the translation to the appropriate IaC provider, whether that’s Terraform, Pulumi or AWS CDK.
AI Changes Everything (Again)
As developers using AI code assistants pile up mountains of pull requests, the challenge becomes ensuring that infrastructure provisioning keeps pace with code production. How does AI code generation fit into Nitric’s IaC model?
“I think when people hear infrastructure from code, they get really nervous that it’s just a developer’s code, and then infrastructure just comes out of nowhere. That’s not true at all,” said Demchuk. “We are actually building guardrails based on your platform team’s specifications, and then that’s what the developers are self-serving from.”
The result is consistently applied security, infrastructure provisioning and architectural decisions, while AI handles the coding work to create the proper deployment.
The result is a compelling workflow: Developers can leverage AI to rapidly generate application logic, while Nitric ensures that the underlying infrastructure is properly configured, secure and follows organizational best practices. It’s automation layered on automation, with each tool handling what it does best.
The Platform Team’s Dilemma
Platform teams, however, can still worry about losing oversight when infrastructure generation becomes automated. Demchuk acknowledged this challenge directly: “I think the biggest thing is we always talk about convenience versus control.”
But, he argued, this concern is misplaced. Nitric doesn’t eliminate platform team governance; it enforces it consistently. When a biomedical customer needs specific serverless architectures for data capture, the platform team defines those patterns once, and Nitric ensures they’re applied correctly across all deployments, whether on AWS Lambda or Google Cloud Functions.
The full conversation dives deeper into customer transition strategies, the technical architecture behind Nitric’s multilanguage support and specific examples of how enterprise customers are using the platform to accelerate their cloud modernization efforts.
For engineering leaders grappling with the complexity of modern cloud development, this episode of The New Stack Makers offers a glimpse into the evolving relationship between developers, platform teams and infrastructure in the age of AI-powered software.
