Skip to main content

Company file

A small technical workshop for workflow agents.

Aboriginal Armadillo is a builder-led agency focused on practical AI systems for work that has become too slow, too manual, or too trapped in one person's context.

Start a Workflow Intake
01Position

Observed before automated.

The work starts by studying what is actually happening: who touches the workflow, which tools carry the evidence, where review belongs, and what a useful first pilot could prove.

02Builder

Built by Trevor Grant.

Trevor Grant is an AI and open-source systems engineer who has worked across production machine learning, agent tooling, developer infrastructure, and open-source communities.

03Principles

How we think.

Start with the workflow

A model is not a system.

Keep judgment visible

People own approvals, taste, risk, money, and professional responsibility.

Build around available resources

Tools, data, permissions, examples, and review loops shape what is possible.

Prefer useful over autonomous

The best first agent is often small, boring, and checked by a person.

04Artifacts

The archive is part of the proof.

Old projects are kept because they show how the work was learned: agent tooling, production ML writing, open-source systems, generative experiments, civic automation, rewrites, dead ends, and useful failures.

Artifact drawer

Prior work preserved

Shows
What was tried, built, rewritten, deprecated, and learned from.
Inspect Artifacts
05Fit

Best conversations start with a real workflow.

Send the repeated task, the tools involved, where it gets stuck, and what a useful first output would look like.