Research

The research behind Myzelium.

Myzelium is not built on claims. It is built on peer-reviewed work accepted at two of the leading venues in software engineering, and validated on a real 50+ developer org. Here is the science, and what it means for how you build.

Two papers. One argument.

Our research answers two questions that matter to any team building with AI.

Question one

Do teams actually need a structured way to work with AI, or is that just tooling overhead?

Our maturity-model work shows the gap is real and measurable: teams move through distinct stages of capability with AI-assisted engineering, and most are stuck early, paying for AI without the structure to get value from it.

Question two

Does a graph-based approach actually solve it in production?

Our meta-framework work shows it does, documented as an industry case, not a lab demo.

Myzelium

Put together: the problem is real and measurable, and the mechanism works in the field. That mechanism is what became Myzelium.

The papers.

Both peer-reviewed · Both public

Accepted at ASE 2026, Munich

SpecGraph: A Meta-Framework for AI-Assisted Software Development, an Industry Case

SpecGraph introduces a way to represent an entire software project as a connected graph of specs, code, and decisions, then use that graph to drive AI-assisted work. The paper documents applying this in a real production software organization, not a controlled experiment, and reports how it changed the way work was planned, executed, and verified.

Why it matters for Myzelium

This is the mechanism behind the product, proven in the field. The living graph you see in Myzelium is the applied form of this framework. The paper is the evidence that it works outside a demo.

Read the full paper

Accepted at ICSME 2026, Bologna

CMM-AISE: A Cognitive Maturity Model for AI-Assisted Software Engineering

CMM-AISE defines a maturity model for how teams adopt and mature in AI-assisted software engineering, describing the distinct stages teams move through and what separates teams getting real value from AI from teams simply spending on it.

Why it matters for Myzelium

This is the evidence that the problem Myzelium solves is real and structural, not anecdotal. It gives teams a way to locate where they are today, and it explains why context and structure, not just more AI, is what moves a team up the curve.

Read the full paper

Developed with academic rigor.

Both papers were developed in partnership with a leading university and passed peer review at top international software-engineering venues. This is not marketing dressed as research. It is research that happens to have become a product.

From paper to production.

The research is the foundation. Myzelium is what it looks like running on your project.

Explore more research at shiftmanifesto.com