Public methods
We do not work with black boxes.
The method is published with the result. Every conclusion can be traced back to the source that supports it.
Welcome to Enpath
Where evidence meets artificial intelligence.
Applied research for governments and organizations that make decisions with artificial intelligence, on evidence anyone can check.
Research · Advisory · Training
Trust
Built on evidence.
-
46
Investigations and analyses published
-
100%
With their methodology published
-
Traceable
Every conclusion, back to its source
-
Human
Final review by a researcher
-
AI + people
The machine analyses; people answer for it
How we work
From evidence to decision.
Five steps, always the same, published alongside every report.
-
01
Problem
A decision with real consequences, framed as a question that can actually be answered.
-
02
Sources
We gather regulation, contracts, official data and literature, and record every one of them.
-
03
AI analysis
Artificial intelligence reads thousands of pages and finds what a human team could not reach.
-
04
Human verification
Our researchers check every claim against its source. Nothing is published without that step.
-
05
Recommendation
Traceable conclusions, with their method, their confidence level and their limits in plain view.
Research
Open research on artificial intelligence and public decisions, with a public methodology and verifiable sources.
See the publications
Government advisory
AI governance, evidence-based decisions, and transformation of the state, for ministries, regulators, and subnational governments.
Services for governments
Private-sector advisory
AI governance and risk and evidence-based strategy, for boards and executive teams.
Private-sector services
Data governance
Data-governance operating models — roles, processes, and standards — as the foundation of any AI or evidence strategy.
About the service
Training
Training for public and executive teams on decisions with data and artificial intelligence.
See the programs
Independent audits
Audits of AI systems with a public methodology and a non-negotiable result — a verifiable external signal.
About the audits
Where we stand
Why Enpath?
Supervised AI
AI speeds up the analysis; people answer for it.
No finding reaches a report until a researcher has checked it against its source and signed for it.
Independence
We do not build AI to sell it.
We evaluate existing technologies. When we make a judgment, we have no product of our own to defend.
Methodology
What a public methodology means.
Five documented stages. You can walk through them here, and find them in every report.
We start with the decision, not the data.
We define what is being decided, who decides, and what evidence would change that decision. A badly framed question is not fixed later with more analysis.
Published: the scope and the research questions
Primary sources, with their provenance.
We identify regulation, contracts, official data and reviewed literature. For each source we record its origin, its date and its known limitations.
Published: the full inventory of sources
AI reads the corpus; nothing is left loose.
Automated analysis processes the whole set and extracts candidate findings. Each one stays tied to the exact passage of the document that supports it.
Published: the assessment criteria
We say how much we know, and how much we do not.
We grade every finding by the strength of the evidence behind it. Where the evidence falls short, we say so, instead of rounding towards certainty.
Published: the confidence level of every finding
Someone puts their name to it.
A researcher checks every claim against its source and takes responsibility for the document. Responsibility is not delegated to a model.
Published: who reviewed it, and when
Research
Recent publications.
Every publication includes its methodology, its sources, and a downloadable version.
Do you need evidence for an important decision?
We help governments and organizations turn complex information into responsible decisions.