- AI infrastructure visualization that once required dozens of engineers and years of work now takes a single developer one afternoon to ship.
- Stripe Projects introduced CLI-based service provisioning, creating a new need for AI infrastructure visualization tools.
- The Stripe Projects Visualizer maps provisioned services and data flows automatically from a project’s state file.
- The speed shift from years to hours is exciting for builders — but raises real questions about job security and security risks.
- AI infrastructure visualization that once required dozens of engineers and years of work now takes a single developer one afternoon to ship.
- Stripe Projects introduced CLI-based service provisioning, creating a new need for AI infrastructure visualization tools.
- The Stripe Projects Visualizer maps provisioned services and data flows automatically from a project’s state file.
- The speed shift from years to hours is exciting for builders — but raises real questions about job security and security risks.
AI Infrastructure Visualization Has Crossed a Threshold Nobody Was Ready For
Something quietly significant happened recently in the world of developer tooling. A single engineer, in a single afternoon, built a working AI infrastructure visualization tool for Stripe’s new Projects platform — the kind of tool that, just a few years ago, would have required a team of half a dozen engineers and the better part of a year to ship. That’s not a marketing claim. It’s a firsthand account from Anna Spies, a developer who has spent the last seven years building exactly these kinds of tools at scale, and who is now watching the effort required to build them collapse in real time.
The story starts in 2018, when Spies joined Stackery, a Portland-based startup that built a two-way editor on top of infrastructure as code. You could drag and drop cloud resources onto a canvas and get deployable IaC back. Or you could write the code yourself and watch it become a live diagram. It was genuinely useful — the kind of tool that made distributed serverless systems legible to teams who weren’t deep in the AWS console every day.
Stackery didn’t survive as an independent company past 2021, but it was acquired by AWS and eventually became AWS Infrastructure Composer, now a full-featured visual designer for serverless architectures using CloudFormation templates. That evolution didn’t happen quickly or cheaply. According to Spies, the initial preview launched at re:Invent 2022 took five engineers over nine months of intense work. The team later expanded to nearly a dozen engineers plus a dedicated UX team, and spent another full year re-architecting the canvas layer to make it more portable and extensible.
What Stripe Projects Actually Changes
Fast-forward to today, and Spies is at Stripe. Not a cloud provider — so, you’d think, no obvious use case for AI infrastructure visualization. That assumption didn’t last long. Last month, Stripe shipped Stripe Projects, a new capability that turns the Stripe CLI into a full provisioning tool. Developers can now spin up services from an expanding catalog covering hosting, authentication, AI providers, and databases — all from the command line.
It’s a meaningful shift in what Stripe is positioning itself to be. The company built its reputation on payment infrastructure, but Projects signals an ambition to become a broader developer platform — not just handling money movement, but handling the scaffolding around entire applications. Think of it as Stripe’s answer to the proliferation of composable, multi-provider app stacks that have become the default architecture for modern startups.
Spies used Projects to build a transcription app for her team, pulling in just two providers: Vercel for hosting and OpenRouter for AI model access. Even with that relatively modest setup, she saw an immediate need for something that would let teammates understand how the pieces connected. So she built the stripe-projects-visualizer — a tool that reads a project’s .projects/state.json file, analyzes how provider environment variables flow through the codebase, and generates an architecture diagram showing which services connect to which.
The whole thing took one afternoon.
The Numbers That Put This in Perspective
It’s worth sitting with the timeline here, because the compression is striking. Stackery — six engineers, multiple years. AWS Infrastructure Composer’s initial release — five engineers, nine-plus months. The canvas re-architecture alone — a full year of additional work. The Stripe Projects Visualizer — one person, one afternoon.
That’s not a marginal efficiency gain. That’s a different category of productivity entirely, and it’s driven almost entirely by the maturity of AI coding tools. Spies herself notes that as recently as eight months ago, she was still struggling to use tools like Kiro and Claude Code to improve Infrastructure Composer’s underlying canvas technology. The tools weren’t quite there. Now, apparently, they are.
The AI infrastructure visualization space has been building toward this kind of democratization for years. What’s changed isn’t just raw model capability — it’s the integration of those models directly into developer workflows, the availability of well-documented APIs like Stripe’s, and the emergence of platforms like Projects that give AI agents structured ways to provision real infrastructure. When an agent can not only write code but also spin up a database or configure an auth provider, the surface area of what one person can build in a day expands dramatically.
The Part Nobody Wants to Say Out Loud
Spies almost titled her post “The Terrifying Reality of Years to Hours,
