Research practice / Praxen Development & Engineering

The question.
The method.
The evidence.

Research on how AI systems are evaluated, governed and developed. The Institute makes room for the question that needs a paper, an explicit model or a reproducible method.

01 / Research programs

Questions under study.

Working paperAgentic Real Options

How should the cost of a deferred AI capability be understood?

Agentic Real Options examines AI technical debt through the lens of options: the cost of waiting, the conditions that change a decision, and the value of preserving the ability to adapt.

The work connects a conceptual framework to engineering choices such as retrieval upgrades and evaluation requirements. Assumptions, costs and decision boundaries need to be explicit enough for others to inspect.

Publication status: a public preprint is not linked here yet.

Research directionSelf-driving laboratories

What makes agent-led experimental work reviewable?

This line of inquiry concerns governance and evaluation when software helps plan, execute and interpret an experiment. What records let another researcher reconstruct the process? Where should approval, intervention and accountability sit?

The focus is the research process and its evidence, including how computational exploration relates to a more costly physical experiment.

02 / Research standard

Make the reasoning inspectable.

State the question: Define the claim, its scope and the assumptions required to evaluate it.
Show the method: Describe inputs, comparison points and resource costs clearly enough for scrutiny.
Separate status from ambition: Distinguish a working paper, a public preprint, a published result and a proposed direction.
Keep a correction path: Make sources, versions and changes part of the record.

03 / Reading and publication

Research and public argument
have different jobs.

Institute papers will be indexed here with their authorship, abstract, version, publication status and downloadable source when available.

For current founder-signed policy analysis, visit Praxen Policy. It is a separate publication of Praxen Development & Engineering, with its own editorial standards and correction record.

Research correspondence

Bring a question worth examining.

Methods, relevant literature and research collaboration inquiries:

institute@cld-dev.io