The answer first.
Plain-language response, key facts, evidence status, and the most important uncertainty.
PROJECT LOOM · PUBLIC ACCOUNTABILITY RESEARCH
Ask a question about a law, policy, public promise, agency decision, program, or political claim. Start with a simple answer. Follow the evidence as far as you need.
Will this legislation pass?
Did the governor keep this promise?
Has this policy actually been implemented?
What happened to the money?
Who was responsible when the outcome changed?
Loom is designed to turn questions like these into evidence-backed research rather than forcing you to become a researcher first.
ASK LOOM
You don't need to understand Loom's research architecture to use it. Ask your question in ordinary language. The public interface is being built to return the simplest useful answer the evidence can support—and to show you what remains uncertain.
Start a question →THREE LEVELS OF DEPTH
Plain-language response, key facts, evidence status, and the most important uncertainty.
Decision, responsible actors, implementation, evidence, contradictions, and unresolved questions.
A deeper research request with source verification, evidence-gap work, competing explanations, and an accountability record.
A Quick Read is not a lower standard of evidence. It is a smaller research surface.
THE QUESTION UNDERNEATH
Policies are often described as though the announcement is the event. But between a decision and an outcome sit agencies, contracts, technical systems, budgets, legal authorities, people, dependencies, information, timing, and institutional capacity. Loom studies that middle layer.
ACCOUNTABILITY
THE LOOM ACCOUNTABILITY RECORD
Loom can reconstruct the chain from a public commitment to the institutions and authorities responsible for carrying it out, then examine where implementation diverged, what evidence establishes that divergence, and whether anyone with authority acted afterward.
Start an accountability question →WHY LOOM IS DIFFERENT
Loom combines primary-source investigation, open research, structured claims, provenance, institutional reconstruction, and mechanical validation around one problem: understanding what happens between political decisions and real-world outcomes.
The interface should be accessible to anyone. The underlying research record should be inspectable by anyone who needs to challenge it.
The system is allowed to say: we don't know yet.
See how the research works →WHAT YOU CAN INSPECT
Plain-language findings for people who need to know what is happening.
Sources, extracted evidence, claims, contradictions, verification, and documented gaps for people who want to check the reasoning.
Machine-readable artifacts, provenance records, schemas, and validators for auditors and researchers.
THE CORE RULE
Loom uses automation where it can improve consistency, traceability, and error detection. But generated text does not become established research simply because it sounds plausible.
Contradictions are recorded rather than smoothed over. Evidence gaps are first-class results. Publication can fail when the underlying record fails the rules.
Inspect the evidence model →