Skip to main content
Heirloom Praxis
Heirloom Intelligence

Published openly, usable without us

What we publish, we publish so it can be used by people who will never hire us. Everything here states its sources, and where we have not done the research ourselves, we say whose research it is.
Published

Available now

framework

The Heirloom Shield: Seven Domains of Sovereignty

A governance framework covering human, data, cultural, economic, creative, community and ecological sovereignty. Published openly, usable without us, and explicitly not a certification.

Sources

  • CARE Principles for Indigenous Data Governance — Global Indigenous Data Alliance
  • NIST AI Risk Management Framework

framework

The Heirloom Method

How we work: harvest what exists, examine it, imagine better, root every decision in people and governance, layer technology thinly, operationalise it, transfer ownership, then multiply.

framework

What an AI Policy Covers

Ten sections, three depths, and what goes wrong when each one is missing. A one-page use guide is not a policy with sections deleted — it answers six of the ten in a line each, and the other four by ordinary practice. Rendered from the same framework the policy builder enforces.

framework

How We Measure, and What We Refuse to Claim

The measurement framework, published before there is anything to measure. Six levels a number can sit at, the alternatives we cannot rule out, and the two claim levels nothing here currently reaches. Rendered from the same registry the code enforces, so the published method cannot drift from the applied one.

essay

The Governance Gap: What the 2026 Adoption Data Actually Shows

Adoption is nearly universal and impact is nearly absent. A reading of published 2026 benchmark data on why the bottleneck is governance and ownership rather than access to tools — with every figure attributed to its named source.

Sources

  • 2026 Nonprofit AI Adoption Report — Virtuous (benchmark study, 346 nonprofits)
  • Unlocking the Promise of AI for Black Learners and Workers — Jobs for the Future
The Dispatch

The newsletter, assembled from the record

Each issue is composed by an algorithm from material this site already holds — when a directory entry was last checked by a person, when a framework was published, how many people ran a tool and what they said took their time. No language model writes any part of it.

Every issue also prints what it could not say. Where the records will not support a section, the section is refused and the reason is published in its place, so a thin month reads as a thin month rather than as a full one.

No issue has been published yet.

The composer runs against a corpus that is still small, and an issue is only published once a person has reviewed it. Nothing is auto-published, and no placeholder issue is shown here in the meantime.

Why an algorithm and not a writer

Because the job is noticing, not persuading. What changed in the directory, what aged out, what people asked the tools — that is arithmetic over records, and arithmetic does not get tired, flatter anyone, or round a number in its own favour. The parts that need judgement are the parts a person still writes, and those are labelled as written by a person.

Not yet

What we have not published

An unusual thing to put on a website, and the most useful thing on this page.

State of Black AI — pilot report

Planned — Year 2

Our first piece of original research. It will not be published until it has a defined sample, a disclosed methodology and outside review, because a report without those things is an advertisement wearing a lab coat. Planned for Year 2.

Why this page is short

A young institution can fill a research section quickly by publishing marketing with citations attached. We would rather have three things that are true than twelve that are impressive. When we have conducted original research — with a defined sample, a disclosed methodology and outside review — it will appear here and it will say all three of those things about itself.