Skip to content
Luiz Hogrefe

AnyLAI

Independent research and development project

Independent R&D project

AnyLAI is an independent research and development project by Luiz Hogrefe, based in Cologne. It is where I develop, test and publish work on governed AI and verifiable evidence.

Visit anylai.eu

What it is

AnyLAI brings together frameworks, runtime research, demonstrations and technical writing in one place, under one discipline: each public statement should lead back to a source, a document or a test.

The maturity of every item is recorded in a single file in the repository, so a prototype cannot quietly become a platform on a web page.

What it is not

AnyLAI is not presented here as a current incorporated company. It sells nothing, has no customers and makes no offer. Earlier commercial activity under related names is historical and is described as such on the Work page.

Focus

  • 01

    Governed AI

    Models doing bounded work under rules that decide, record and can be replayed.

  • 02

    Evidence

    Values that carry their source, their confidence and the person who reviewed them.

  • 03

    Semantics

    Vocabularies and data models that keep their meaning across systems and languages.

  • 04

    Interoperability

    Public standards for passports, credentials and identifiers, rather than private formats.

  • 05

    Product data

    Digital product passports under European product law, built from evidence.

  • 06

    Decision architecture

    How a runtime qualifies what to do next and what it is allowed to do.

What anylai.eu publishes

  • COADF 2.2

    The framework, its principles, conformance manifest and a non-normative Engineering Companion.

  • ERQYO

    The runtime architecture and the status of its reference implementation.

  • Research concepts

    One page for each applied prototype, with its question and its limits.

  • AnyDPP and AnyAudit

    A research and demonstration environment and a research demonstrator, each on its own domain.

  • Writing

    Articles on AI governance, product passports, evidence and European regulation.

  • Regulatory Radar

    A dated record of regulatory changes relevant to the research.

How it works

The framework is applied to the project that produced it. COADF's conformance run executes against AnyLAI's own repository, and fences in the pipeline stop public copy that asserts legal conformity or an unverified fact.

Where a statement cannot be supported yet, it is labelled as a gap rather than rounded up.

What lives where

hogrefe.ai

The person: career, working principles, an overview of the research and where to look next.

anylai.eu

The laboratory: frameworks in full, research details, technical demonstrations and regulatory research.

  • COADFPublished framework

    A published framework for building AI-assisted software under explicit governance: eight principles, a control manifest and CI-enforced fences.

    COADF

  • ERQYOReference implementation · not deployed

    A runtime architecture that lets learned rhythm shape a proposal while authoritative constraints decide the operation.

    ERQYO

  • Trust ArchitectureImplemented platform architecture

    Provenance, evidence, confidence, authority and audit, arranged so that trust can be checked at every boundary.

    Trust Architecture

  • Open Template LibraryPilot catalogue · prototype

    A pilot catalogue of implementation templates, each written once as a source and generated for machines, AI tools and people. Five are published.

    Open Template Library