E199: Rethinking Infra-as-Code for the AI Era

The podcast discusses the founders' extensive backgrounds in IT and operations, highlighting their experience in building and scaling complex systems across industries. Their work at Instana, an APM company acquired by IBM, revealed significant challenges in monitoring microservices and managing infrastructure at scale. These frustrations led to the creation of Platform Engineering Labs, focused on simplifying infrastructure management through automation and addressing inefficiencies in current DevOps practices.

A central theme is the critique of existing Infrastructure as Code (IaC) tools like Terraform and Pulumi, particularly around state management, configuration drift, and the "blast radius" of changes. The discussion introduces Formae, a new IaC platform designed to solve these issues with a hybrid declarative language called Pickle, which offers schema safety and programming constructs without full-language complexity. Formae emphasizes an agentic backend for automated reconciliation, ephemeral patching, and a two-loop system that separates regular updates from immediate, temporary fixes.

The podcast also explores broader industry challenges, including the slow adoption of open source in SRE/DevOps circles, the need for better abstractions to support AI-driven infrastructure, and the limitations of current tools in handling AI-generated changes. The founders advocate for a shift toward automated, code-based infrastructure that minimizes manual intervention and human error, while promoting open source to foster trust and community-driven improvements. The platform supports gradual adoption, integrates with existing tools like Terraform, and offers a free, unlimited core edition to lower barriers to entry.

• What if you automated discovery and patching of infrastructure drift in real time?

  • Move: Use an open-source tool like Formae to automatically discover your live infrastructure, generate up-to-date IaC code, and allow ephemeral patching to fix drift without full redeployment.
  • Why Now?: Drift is compounding as more changes come from autoscalers, humans, and AI agents; manual reconciliation is unsustainable for solo developers managing growing systems.
  • Expected Upside: Reduce debugging time by 50%+ and eliminate surprise outages caused by configuration drift, while maintaining full code-reviewed workflows for permanent changes.

• What if you managed infrastructure with a free, scalable IaC tool that doesn't charge per resource?

  • Move: Replace or complement your existing Terraform/Pulumi setup with a core-free IaC platform (e.g., Formae) that offers unlimited resources in its free tier and charges only for enterprise scaling features.
  • Why Now?: Rising infrastructure complexity makes per-resource pricing models cost-prohibitive even for inactive components; solo operators need zero-cost baselines to stay lean.
  • Expected Upside: Cut monthly tooling costs by up to 100% on small-to-mid projects while gaining access to advanced backend capabilities like automated sync and patching.

• What if you integrated AI agents safely into your IaC workflow without risking production?

  • Move: Implement a two-loop system - one for standard code-apply (reconciliation) and another for ephemeral, minimal AI-generated patches - using a platform that treats all changes as code, even temporary ones.
  • Why Now?: AI copilots are generating risky, oversized PRs; solo developers need safe abstraction layers so AI can assist without causing blast radius failures.
  • Expected Upside: Accelerate infrastructure iteration by 30 - 60% using AI while maintaining human oversight, reducing both toil and risk of catastrophic errors.

• Audit your current infrastructure for undocumented or forgotten resources using automated discovery tools to uncover hidden technical debt.
• Implement ephemeral patching workflows to make small, temporary infrastructure changes without risking broad system disruptions.
• Adopt a modular Infrastructure as Code (IaC) approach that avoids holistic apply cycles, reducing blast radius by isolating change impacts.
• Use a schema-safe, hybrid declarative language (like Pickle) instead of raw YAML or HCL to prevent configuration errors and improve code reliability.
• Launch with a fully functional open-source core to build trust and community adoption, reserving paid features for scalability and enterprise needs.