We built a governance layer for AI-assisted development (with runtime validation and real system)

Dev.to / 3/28/2026

💬 OpinionDeveloper Stack & InfrastructureSignals & Early TrendsModels & Research

共有:

Key Points

The article describes Janus, a governance layer for AI-assisted development that focuses on validating protocol conformance and producing governance evidence rather than measuring only model performance.
It highlights progress in a second paper that turns a theoretical governance model into measurable components including evidence-based scoring, omission detection, human authority boundaries, and deterministic reconstruction.
The team provides benchmark metrics (ECR, GVL, PVDR) to evaluate the governance approach and compare outcomes in a structured way.
A live system is running on top of the framework, with additional resources shared for the validation framework and the public demo.
The author asks for feedback specifically from experts in observability, event sourcing, and AI tooling, indicating an emphasis on real-world integration and tooling feedback loops.

I’ve been working on a project called Janus — a governance layer for AI-assisted development systems.

The core idea is simple:

Instead of evaluating performance, we evaluate governance through evidence and protocol conformance.

We just published the second paper, which moves from the theoretical model to measurable governance:

Paper 1 (model):

Paper 2 (validation):

There’s also a live system running on top of it:

And the framework used to run it:

I’d really appreciate feedback — especially from people working on observability, event sourcing, or AI tooling.

This article is featured in our daily AI news digest — key takeaways and action items at a glance.

AI Business

Dev.to

Dev.to

Dev.to

Dev.to