A laboratory for AI systems and the engineering behind them.

Why we exist

Artificial intelligence has moved from research prototypes to systems that people rely on for decisions that matter. The gap between what those systems appear to do and what they can be shown to do is where most of the risk lives. LAISER was founded to work in that gap: to study how AI systems are built, explained, evaluated, and operated, and to publish methods that others can hold their own work to.

What we do

We run research programmes in foundation models and language systems, explainability, multi-agent and agentic systems, safety and evaluation, applied data science, ML systems engineering, geospatial intelligence, and long-horizon forecasting. The programmes share one methodology, a common infrastructure, and a habit of writing down what did not work.

Our applied work currently centres on land and development markets, a domain with large, fragmented, time-dependent data and ground truth that eventually arrives. It is a demanding test bed for everything above.

How we work

Everything is in version control. Research code, data registries, experiment reports, and this website live in our GitLab group, with protected branches, automated checks, and pipeline-only deployments. We use AI coding agents extensively, under an operating contract that forces them to label evidence, test for leakage, and submit to adversarial review, and we publish that contract as a framework.

Non-profit status

LAISER operates as a non-profit research laboratory. Our research outputs are published for the benefit of the field; our code is released under licences chosen per project.