Architectural Coherence Index
Your codebase has a trajectory. Now you can see it.
The Architectural Coherence Index is a structural diagnostic for codebases built with AI coding tools. It measures drift before velocity collapse.
The problem
Velocity up. Confidence down.
Your team is shipping faster than ever using Claude Code, Cursor, or Copilot. The builds are green. The dashboards are clean. And something still feels wrong.
AI coding tools amplify whatever governance level your codebase already has. In some codebases, they enable cleanup and refactoring. In others, they concentrate dependencies on a small number of hub modules in a way that gets harder to undo every month. The difference is measurable. It is also invisible to every standard tool in your stack.
Most teams find out which side of that distribution they're on the hard way: a slow refactor that nobody can finish, a hire who can't onboard, a feature that takes six weeks instead of two.
You don't have to wait for that.
What we found
The Modular Mirage is real and detectable.
We've been running a longitudinal study on dependency graph structure across Python open-source repositories, tracking topology before and after documented AI tool adoption.
The original hypothesis was that AI tools uniformly increase coupling. The data rejected it. What the data actually showed was divergence. Some codebases concentrate dependencies on hub modules post-adoption. Others diversify away from them.
In the codebases that drift, the structural signature has a name in the recent literature: the Modular Mirage. Clean folder structure. Many small files. Reasonable test coverage. And underneath, a handful of modules absorbing more of the system's dependency weight every month.
Super-node modules become review bottlenecks. They resist decomposition. Changes to them are high-risk and high-cost. The system looks healthy until those modules become impossible to touch. Then the velocity collapse feels sudden. It was not sudden.
By the time the collapse arrives, the structure has been telling you for months.
Who this is for
Engineering leaders who feel velocity-up confidence-down and want to know if it's real.
Fractional CTOs who need a diagnostic that converts discovery calls into paid advisory engagements.
Founders preparing for fundraising, acquisition, or enterprise customer scrutiny.
Investors assessing technical risk in portfolio companies before term sheets.
Not for: teams looking for general code quality metrics. CI integration. PR feedback loops. Security scanning. Those exist. This is a different category.
The closed beta
ACI is currently in closed beta with selected fractional CTOs and engineering leaders. The beta program includes:
- A full diagnostic run on one repository of your choosing.
- A written executive brief with top three risks and a prioritized roadmap.
- A 30-minute walkthrough call to interpret findings and answer questions.
- Direct input into the product roadmap and pricing as it moves out of beta.
We're being selective about beta participants because the diagnostic depends on direct repository access, clear engagement context, and your time. Spaces are limited and we're prioritizing applicants who will use the output to drive an actual decision.