§ Research

The science behind SHIFT

SHIFT is grounded in peer-reviewed research on how AI changes the way engineers think, learn, and build. Papers from the founding organizations are collected here as they are published.

Publications 2 available · 1 under review
Preprint IEEE ICSME 2026

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

D.A. González-Rueda, M. Linares-Vásquez, D. Soto, A. Hurtado

Uniandes · Tres Astronautas · Blend 360  —  To appear, 2026

Argues that AI-assisted engineering creates a tension between cognitive scaffolding, which strengthens reasoning, and cognitive substitution, which bypasses it — and proposes a maturity model spanning capability dimensions, maturity levels, and stage-dependent risk across the SDLC.

Read the paper
Preprint IEEE VISSOFT 2026

Re-Imagining Software Visualization for Spec-Driven Development

G. Rosa, M. Linares-Vásquez, G. Robles

Universidad Rey Juan Carlos · Uniandes  —  To appear, 2026

Specifications are now first-class artifacts, yet they are still handled as flat text — the same comprehension problem that motivated code visualization two decades ago. Proposes a five-level framework for visualizing SDD projects, from the structure of a single spec through to traceability into generated code and its coevolution over time.

Read the paper
Under review

Spec-Graph Meta-Framework for AI-Assisted Software Development: An Industry Case

Tres Astronautas · Uniandes

An industry case for representing system specifications as a typed graph, so that AI agents consume structured, traceable context instead of loose prose. Presented as a talk at the SHIFT Forum 2026. The paper is under review and will be posted here once a decision is in.