aiclavis
We test what the evidence actually supports — including the possibility that the suspected problem is not there.
Core principle
aiclavis research is not designed to confirm an initial hypothesis. A research engagement is a success when the evidence supports the original concern — and equally a success when it shows the suspected gap is not there.
The objective is to resolve the question, not to validate the concern.
The objective is to close the question — in either direction.
Method
Every conclusion in aiclavis research is marked by what kind of support it actually has.
Directly supported by the available evidence.
Calculated or logically derived from observed evidence through an explicit method.
An explanation supported by the evidence, but not directly observed.
The available evidence is insufficient to support a conclusion.
Independent research
A $400M project is not necessarily a $400M decision: whether it reflects a discrete order or a broader manufacturing-footprint decision already forming across the portfolio.
View research →Same physical volume, different economic regime: a 2Q26 3.7× EBITDA-per-tonne divergence between Gerdau's Brazil and North America segments.
View research → View all research →What aiclavis investigates
What is actually committed, optional or already constrained?
Where do reported outcomes conceal materially different economic states?
Which assumptions, dependencies or commitments narrow the available decision space?
Does the evidence actually support the concern — or should the hypothesis be closed?
Work with aiclavis
Define the decision question, evidence boundaries, falsification criteria, required people/data, and the value at stake.
Tests a specific decision question against the available evidence. The result may confirm, narrow, reframe, or close the hypothesis.
Founder & Research Lead
Nika Novak is an independent researcher and founder of aiclavis. Her work focuses on reconstructing decision environments from fragmented evidence, with an explicit separation between observation, derivation, inference and what remains unknown.
About Nika Novak →Research integrity
aiclavis research distinguishes observed evidence from calculation, inference and uncertainty. Hypotheses may be revised or closed when the evidence does not support them. Material factual corrections are documented.
Companies and subjects of public research are invited to provide factual corrections or official context.
Evidence Standard →If the decision matters enough that "probably" is not a sufficient answer, we can examine what the evidence actually supports.