PRINCIPLES

What REGULOGRAPH is — and what it deliberately is not.

The platform's boundaries matter as much as its capabilities. Below is an explicit statement of what REGULOGRAPH is responsible for — and what it is not.

01Framework

Is / Is not

A clear delineation of what the platform does and what it deliberately does not.

Is
  • A system for managing regulatory-sensitive claims, statuses, evidence, approvals and publications.
  • An auditable regulatory memory of the company and a timeline of interactions.
  • A tool for prevention, remediation, re-checks and preparation of records.
  • A source of the structured Evidence Pack.
Is not
  • A regulator, licence, certification, safe harbour or guarantee of compliance.
  • A substitute for qualified legal advice.
  • An autonomous decision-maker; externally significant legal steps are approved by a named person.
  • A predictor of the outcome of a regulatory, administrative or judicial process.
  • A tool for monitoring individuals or attacking individual officials.
02Principles

Eight principles we build by

The platform's construction rests on firm principles, not ad hoc features.

01

Evidence before scores

The evidentiary basis takes precedence over any aggregate rating; without a source, a claim is not valid.

02

The claim is the basic unit

We work with atomic claims, not whole documents, because risk arises at sentence level.

03

Scope is mandatory

Every claim has an entity, product, jurisdiction, audience, channel and time period; without scope, nothing is published.

04

Primary sources prevail

Official texts, decisions and guidance outweigh secondary interpretation and opinion.

05

Uncertainty is displayed

When a claim cannot be confirmed, the system abstains from a conclusion and marks it as unsupported.

06

A person carries the name and the responsibility

Every significant decision has a named owner and approver; automation does not replace accountability.

07

Silence is not consent

The absence of a reply or correction is never construed as consent, a licence or a promise.

08

Full export

All claims, sources, approvals and interactions can be exported at any time in human- and machine-readable form.

03AI governance

How AI works in REGULOGRAPH

AI proposes; a human decides. Deterministic rules take precedence for blocking wording.

  • Proposals always carry a citation and the date of the source.
  • Deterministic rules take precedence for blocking wording.
  • A mandatory confidence threshold and abstention — when a claim cannot be confirmed, the system withholds a conclusion.
  • Client data is not used to train models by default.
  • Every significant output is linked to the model, the sources and a human decision.
  • Final approval is always given by a named person.
04Security and evidentiary integrity

Isolation, immutability and controlled access

Evidentiary integrity rests on isolated workspaces, encryption and an immutable audit log.

  • Isolation of customer workspaces.
  • Data encryption.
  • Separate zones for privileged material.
  • An immutable audit log.
  • Controlled export with audience review.
  • Hashes and time stamps prove integrity and time — not the truth of the content.
STATUS
The REGULOGRAPH platform is under development. Core capabilities are detailed to partners in an introductory consultation.
Entry serviceRegulatory Memory ReconstructionContinue →
Made with AI in Macaly